"Talks a good game about freedom when out of power, but once he’s in – bam! Everyone's enslaved in the human-flourishing mines."

Links 10/17: Martin LuthURL

Ulpian’s life table is the closest thing we have to actuarial data from ancient Rome. Key quote: “Although Keith Hopkins called the table not ‘demographically possible’, it corresponds well to other observed populations with abnormally high mortality rates (such as postwar Mauritius), and to a priori constructions of plausible Roman age structures. In any case, the picture they present is appalling: a society with one of the highest mortality rates on record, with a predicted life expectancy at birth of between 19 and 23.”

Vaughan Bell of MindHacks on good and bad criticisms of psychiatric diagnosis. Interesting to me for the “Psychiatric diagnoses are not reliable” section, which shows that although you can’t get reliable diagnoses out of ordinary practice, you can get them from structured interviews. Which no one does.

Chinese censorship expands to group chats in response to dissidents turning them into “lecture halls” for banned material. Related: China cracks down on Winnie-The-Pooh for purported resemblance to Chinese president Xi Jinping.

Independent: “each day, a cruise ship emits as much particulate matter as a million cars. They say that “30 cruise ships pollute as much as all the cars in the UK”, but I wonder if that’s true – eg if particulate matter is the relevant kind of pollution, and how much pollution in the middle of the ocean compares to in major cities. Probably can’t just ban cruise ships and then stop worrying about car pollution so much.

If you like your questionable science served with a side of other questionable science, here’s MBTI Facial Phenotypes, the average face appearance for each of the Myers-Briggs types (eg “ESTJ”, “INFP”, etc)

Man accused of being D. B. Cooper, the famous hijacker who held an airplane hostage, parachuted out with the money, and was never seen again.

Investigative reporter finds that drug offenders sent to “rehab” instead of prison often end up in privately-owned work camps with a thin veneer of rehabness, where they’re exploited as slave labor in terrible conditions. The camps claim that labor can help teach people good work values which prevents drug relapse, but the “programs” are essentially indistinguishable from difficult work in dangerous conditions without pay or protections. Many “clients” end up just choosing to go with the prison sentence after a while. ACLU vowing to investigate.

Speaking of ACLU cases: city of Dickinson, Texas, tries to make hurricane relief aid conditional on recipients promising not to boycott Israel; appears to be attempt to comply with state law banning officials from doing business with Israel-boycotters.

Inspired by my map of online rationality, Søren Elverlin makes a map of the AI safety community. But how could they represent Roman Yampolskiy with a medieval looking building? Shouldn’t they have used something Romanesque? So disappointed.

Not the Onion: Universe Shouldn’t Exist, CERN Physicists Conclude. Not a moral claim, just a study failing to find any asymmetries separating matter and antimatter.

Profile of leading self-driving car researcher Anthony Levandwoski, who used to lead Waymo, co-founded self-driving truck company Otto, and now…is trying to start some kind of church relating to the Singularity?

A cautionary tale: the archives of fuckedgoogle.com, an early 2000s website arguing everyone else was idiots for not realizing Google was a fad that was about to collapse. Useful vaccination against taking confident-sounding people seriously.

Research in Journal of Personality and Social Psychology: People do better than chance at matching people’s faces with their names, suggesting it really is possible to “look like a Bob” or whatever, “even when ethnicity, age, and other socioeconomic variables are controlled for”. Authors claim “people alter their face to meet cultural expectations”, but my guess is it’s going to be some kind of sub-sub-ethnicity sort of thing where Scottish people and Swedish people are both classified as “white” but look sort of different and give their kids different names. Although the studies were done in France and Israel and I don’t know the ethnic situation there.

“My current pet theory of the function of REM is that it is doing the same thing as experience replay in DeepMind’s reinforcement learning algorithms.”

Current Affairs: The Sad Spectacle Of Cities Groveling To Amazon.

Biggest test of police body cameras so far finds they have no effect on police brutality or citizen complaints. Article suggests this is shocking but I remember seeing other studies showing the same last year (but see here).

Puerto Rico is still a disaster area. One bright spot: Google successfully restoring some communications with stratospheric Internet balloons, Tesla successfully restoring some power with portable solar arrays. Meanwhile, in the actual government…

Related: it’s easy to get the impression from the media that everybody hates tech companies, but actually Google’s approval rating is 88%, Amazon’s is 72%, and Facebook’s is 60%. Key quote: “The campaign against big tech isn’t resonating, because it’s based on false premises that most people see right through.”

You know that story about how a neural net trained to detect tanks actually just ended up classifying sunny vs. cloudy days because all the tank pictures were taken during sunny weather? Probably an urban legend.

BBC on Viking re-enactors. Interesting both because Vikings are inherently interesting, and because of the attempt to make a bunch of ‘combat veterans and former football hooligans’ waving war-axes at each other sound touchy-feely and lovey-dovey. The title is “The Viking Therapy Club” and key quotes are eg “According to Qanun the true Viking message, which the Jomsborg Vikings try to promote, is one of tolerance and diversity.”

Related BBC article: Viking textiles have the word “Allah” on them, demonstrating strong connections between Viking and Islamic cultures. Related Twitter followup: No they don’t, although contact between Vikings and Muslims was real enough. A good reminder that everything looks kind of like Arabic if you stare at it hard enough.

Eliezer Yudkowsky on the MIRI site: There’s No Fire Alarm For Artificial General Intelligence. There will be no particular event that creates common knowledge that it’s okay to say AGI seems near, so a conventional wisdom that it’s certainly a long way off will last long past the point when the evidence suggests otherwise.

Related on the new AlphaGo Zero: “I emphasize how all the mighty human edifice of Go knowledge, the joseki and tactics developed over centuries of play, the experts teaching children from an early age, was entirely discarded by AlphaGo Zero with a subsequent performance improvement. These mighty edifices of human knowledge, as I understand the Hansonian thesis, are supposed to be the bulwark against rapid gains in AI capability across multiple domains at once. I said, ‘Human intelligence is crap and our accumulated skills are crap,’ and this appears to have been borne out…[AI proved] fast enough to blow past centuries of human-style learning in 3 days”. And Robin Hanson’s response.

Related…ish: Paperclip maximizer, the game. As if all of you haven’t already played this. If you don’t get the reference, this article explains. Hint: don’t do what I did and use negative quantum operations to turn back time, it just makes you start over from the beginning.

Also: Eliezer is writing a book on the idea of low-hanging fruit vs. the argument from humility (“Surely I’m not so great that I can discover low-hanging fruit everyone else has missed”). First chapter up here, some responses here and here.

The only time you’re allowed to sign up for Obamacare insurance policies this year is between November 1 and December 15 (slightly longer in some states). This is poorly advertised possibly as a sinister Republican plot. If you’re an American and you need health care, check out the Obamacare FAQ sometime during that period so you don’t get left behind.

Relevant to my interests: Ninth Circuit Court rules that whales count as fish for certain legal purposes.

Trump Supporters Help Fill Republican Party Coffers. Key point is probably this graph – whatever other problems Trump is causing the GOP, he’s caused an almost unprecedented flood of small donor fundraising and given the Republicans a major financial advantage going into the midterm elections. May be used for intra-Republican primary conflicts to support Trumpist candidates.

Some good (and not so good) reactions to my post on New Atheism, including Siderea, Nathan Robinson, Jerry Coyne, and the subreddit comment section.

Michael Huemer: What’s Wrong With Soliciting Letters Of Recommendation? For one thing, these have become an irritating exercise in social obligation – most teachers don’t want to ruin their students’ lives or get in trouble, so if the student is at all okay the teacher feels obligated to call them “the next Einstein” or whatever. But it’s also the essence of “who you know rather than what you know”, reinforces the credentialist system, and distributes positions to people who are most sycophantic, most willing to pester others for favors, and who have the teachers most willing to be dishonest. Another factor reinforcing credentialism and undermining meritocracy. Also costs professors many hours of unpaid labor. Huemer recommends weighing other factors instead.

Documents Banned By Section 58 Of The Terrorism Act 2000. British censorship has intensified, with long jail sentences for anybody who reads “terrorist content” online, with “terrorist content” being anything from an ISIS magazine to (potentially) news articles about terrorism. Already used to convict a British Muslim whose brother had gone to fight for ISIS and who had looked up some ISIS stuff to try to understand how her brother was doing. Another person given suspended sentence for possessing a terrorist magazine with a copied Buzzfeed listicle about evading drone strikes.

On how police unions can either fight or reinforce corruption: “The things cop unions do that reformers don’t like – reflexively defend all officers in all situations, fund legal defenses and media campaigns…those are all felt, by cops, as safeguards against police corruption”.

This year’s best Halloween costumes so far: Winnie-the-Pooh’s Tigger (and other variations on a theme), and Person Getting Deported By Trump

This entry was posted in Uncategorized and tagged . Bookmark the permalink.

Leave a Reply

428 Responses to Links 10/17: Martin LuthURL

  1. Bellum Gallicum says:

    Curious about that Roman chart, Ulpian’s table? Certainly doesn’t correspond to my readings of the ancient world, but I wonder what people think it represents?

  2. Bellum Gallicum says:

    I could be reading the source incorrectly but it seems to me that Ulpian’s table is an annuity schedule rather than a tax schedule and seems to imply a life of 19 to 23 from the persons current age which is quite different than total life span.

    • Eric Rall says:

      My read is that it’s a schedule for valuing annuities for the purpose of administering an inheritance tax. I agree with your reading that it implies a life of 19-23 years for the person’s current age.

      A few other things come to mind in terms of interpretation. One is that the table was generated around the beginning of the Crisis of the Third Century, which wasn’t a particularly great time to be a Roman. Another is that the table might be based particularly on urban populations, which tend to be more legible and of more interest to governments, but which have very different demographic characteristics than the broader rural population: pre-industrial cities tend to be people-eating monsters, where the bulk of the population dies off (mostly from disease) faster than they can breed, and thus rely on immigration from the countryside (the so-called “surplus population”) to maintain or increase population. On the other hand, the table might actually overstate life expectancy, since the poor people most at risk from dying of disease and malnutrition are relatively unlikely to be the beneficiaries of annuity-structured inheritances worth taxing.

  3. Douglas Knight says:

    Cruise ship pollution:

    Particulates are the relevant type of pollution.

    Here’s an order of magnitude comparison. The Mediterranean is about 10x the area of the UK. So if there are 300 cruise ships in the Mediterranean, they pollute it to the same extent as cars pollute the UK. The typical point in the UK, ie, a rural location, is not very polluted. However, typical cruise stops should care.

    • t mes says:

      300 cruise ships polluting the Mediterranean to the same degree as 66 million people pollute the UK sounds really bad. Anyway you slice the value per unit pollution is going to come out making cruise ships look like a horrible proposition if the above fact is true.

      • baconbacon says:

        The relevant comparison isn’t cruise ship to car, its cruise ship to household. A ship is providing in addition to transportation heating/AC, electricity, restaurants, bars, gyms, movie theaters etc. If half of all pollution comes from cars then its 60 cruise ships, not 30.

    • bean says:

      Particulates are the relevant type of pollution.

      Shouldn’t this be “Particulates are a relevant type of pollution”? The comparison smells like cherry-picking. If cars emit relatively little particulates relative to their overall pollution profile, and cruise ships a lot, then this doesn’t necessarily tell us much about their overall environmental impact. How about comparing cruise ships to coal-burning powerplants?
      Also, is anyone else in favor of firing up an old marine steam plant to show these people what real pollution is?

    • Steve Sailer says:

      What are “particulates?”

      Are they poison or are they, basically, dust? If the latter, do they sink to the bottom of the Mediterranean, or do they mess it up in some fashion?

      • Aapje says:

        Very light dust that is so small that it is poison (or more accurately, easily goes into your lungs and bloodstream). Because it is so light, it can travel enormous distances. For example, Saharan dust regularly ends up in Europe, in large amounts (so much that it can perceptibly color the sky).

  4. Frank Lantz says:

    Scott, just out of curiosity, did you hit the confirmation box on Quantum Temporal Reversion, or was this before I added it? (Trying to calibrate how bad to feel about this.)

    • Scott Alexander says:

      Before you added it. I couldn’t figure out how quantum computing worked, ended up with a bunch of negative ops, and guessed that the temporal reversion would revert it to zero ops. Don’t feel bad; game was great.

      • Error says:

        Related feedback: I liked the quantum computing mechanic. It was extremely useful, but took experimentation and attention to figure out how to use it reliably.

    • Edward Scizorhands says:

      I would really like to have something explaining what the buttons do. I get that this might be part of the game, and you don’t want to explain everything, but “how much wire does clicking ‘buy’ put into my inventory?” is something I should be able to tell directly.

      Also, maybe some way to disable the autoclippers. I bought two and they’re really just saving me from clicking which I shouldn’t bother doing early in the game.

    • Incurian says:

      I played this all day yesterday. It took me eight hours. I suspect someone smarter could have done it much faster.

      • Error says:

        I did that last night, and then did it again to see the other ending. Something similar happened with A Dark Room when it came out. I think I need to not touch this genre again ever.

        …but it was a good game.

  5. AdamM says:

    I have noticed that some national groups clearly have a certain look that I would strongly bet is recognizable much more often than chance, and it’s clearly not due to genetics. The most obvious examples that I’ve noticed are Russians and Israelis, who in both cases seem to have a certain stereotypical look. There are probably others that I’m just not as familiar with. I don’t know about a correlation with first names though.

    • Scott Alexander says:

      I feel like I can tell white English people from white Americans in a way that doesn’t seem to depend on clothing. Not sure how this could possibly be true.

    • Steve Sailer says:

      I saw a documentary over the weekend about the famous chimp scientist Jane Goodall as a young woman. She was exceptionally English-looking. Her mother looked just like Margaret Thatcher when she was prime minister, another extremely English-looking woman. (Thatcher had been famously good-looking in the 1950s for a young parliamentarian according to several English Tory celebrities, such as Alec Guinness and Kingsley Amis, although they specified that her English looks might not appeal abroad.)

      Here in America, you don’t see very many people as English-looking as Goodall or Thatcher. Whether it is due to more mixed genes or a different set of facial expressions in American culture, I couldn’t say.

    • Creutzer says:

      and it’s clearly not due to genetics. The most obvious examples that I’ve noticed are Russians and Israelis

      Those examples are not going to convince me of “not due to genetics”. Not at all.

      • Wency says:

        Indeed, the point about Israel is just utterly confusing to me.

        Is there a suggestion that Ashkenazim don’t often have distinctive physical traits compared to other white people? That if we didn’t know who he was and just had a photo, we’d have no way of guessing that Jerry Seinfeld is more likely Jewish than Swedish?

        Or are we comparing Jewish Israelis to Jewish Americans? In which case the explanation is that Israelis are largely non-Ashkenazi; Mizrahi Jews are much more darkly complected than most white Americans.

        Or are we comparing Ashkenazi Israelis to Ashkenazi Americans? In which case the Americans are probably admixed more with white gentiles and the Israelis with non-Ashkenazi Jews. Or even with different white gentile populations from the Americans.

        • AdamM says:

          I don’t really think that it’s specifically Ashkenazi Israelis, although I haven’t paid much attention to checking.

          I understand that there are reasons why the genetic pools of Israelis and American Jews are different to some degree, but I just don’t think that that is the explanation of the phenomenon. If it were a case of differing average genetics I would expect that a small but significant proportion of American Jews would look to some degree like Israelis, and I don’t think that’s the case.

          Likewise with Russians, which I think is an even more pronounced example of a clear national “look”. Russian Israelis quite often have the Russian look, but their children who grew up in Israel do not.

      • Jaskologist says:

        Compare American-born Chinese and fresh off the boat Chinese immigrants for a much better example.

        • Steve Sailer says:

          Chinese people who had the misfortune to grow up under Mao learned to adopt a perpetual scowl known as “class struggle face.”

        • Wency says:

          Indeed, I think suntzuanime below has it — a combination of grooming and genetic factors, with the relative weights having to do with the specific example.

          Though even among Chinese-American populations, there’s the possibility of genetic factors contributing (though in your example, grooming clearly dominates). Earlier Chinese immigration to the U.S. was apparently primarily Cantonese; later immigration less so. In a lot of physical characteristics Northern Han tend to differ from Southern Han, including prevalence of shovel-shaped teeth.

      • AdamM says:

        Those examples are not going to convince me of “not due to genetics”. Not at all.

        Because the genetics are so often similar between Israelis and Israeli Russians, it seems unlikely that the different looks are due to genetics. Likewise Israeli Ashkenazim and American Ashkenazim should have basically identical genetics as a group, so I wouldn’t think that the existence of a separate look for Israelis would be due to genetics.

        • Null42 says:

          It’s funny, but I wonder how much of the differences between American and Israeli Ashkenazim is due to self-selection. It’s an interesting example of ceteris paribus: take the same ethnic group and offer them two different environments.

          You either want to serve in the army or you don’t, and you’re either comfortable being a majority in a country that could face annihiliation or a minority in a country that’s extremely unlikely to be annihilated but might decide to annihilate *you* if worse comes to worst (which I also think is extremely unlikely, but after Weimar I can see why Jewish people would be nervous).

          I wonder how much of the difference is due to this? Anyone with more experience care to comment? (I lived in NYC for about 20 years but am not very observant.)

    • suntzuanime says:

      There are probably major cultural differences in hairstyling, makeup use, facial tattooing, dental care, etc.

      Although I wouldn’t be too quick to rule genetics out.

    • Watchman says:

      I wonder about the role of language here – how you think influences how you act, but presumably this also might influence how you communicate non-verbally, and the language you think in would therefore send non-verbal cues. The difference between Americans and the British (living in the UK, I would be loath to try and spot the difference between the various nations in the UK, and the Irish) could be accounted for this way as well – the two Englishs are similiar, but have some notable differences in vocabulary and assumptions (Americans also tend to have wider functional vocabularies in my experience – although the only evidence for this I can find is in the vastly superior rhymns that your average American rapper uses over an average UK rapper, which is clearly a value judgement). That it is perhaps more difficult now than in the 1950s to spot the difference between an Englishwoman and a female American (there must be a correct adjective here but I’m stumped) might reflect the recent convergence of language that is going on.

      This could also explain the fact that people of different social backgrounds (class if you want, but it always feels a bit Marxist) can be identified in the same way.

      Or it could all be confirmation bias on all our parts – which empirically might be the best solution (and certainly the hypothesis I would seek to prove first).

      • LewisT says:

        Americans also tend to have wider functional vocabularies in my experience

        As an American who has had the chance to talk to British people, and British-educated people, on several occasions, my experience has been the exact opposite. Heck, even the actors and comedians on British TV shows seem to have a much larger and more sophisticated working vocabulary than their American counterparts.

        • Joseftstadter says:

          As an American who travels to the UK often, I agree with you. British people tend to value verbal wit and word-play more than Americans do. Verbal facility in America is often perceived as a sign of insincerity and dangerous cosmopolitanism. The last two Republican Presidents have even made a virtue out of verbal incoherence.

          • Watchman says:

            Hmm. Interesting difference of opinion here. Although my actual point was poorly expressed – the reason I cited rap was that I was talking about the basic level of vocabulary rather than that of the educated (speaking as an educated Briton, I am pretty certain that there is minimal difference in lexical knowledge between university-educated native anglophones globally as the education is a leveler). At lower levels of educational attainment, I find it hard to believe that a typical representative of the UK has a wider vocabulary than their US equivalent (apart from probably swear words – we’re good at those as a nation), and I have travelled in the US myself. Mind you, there is a bias for travellers I suppose – you meet people working in service industries, who are selected for ability to communicate.

          • Joseftstadter says:

            In London at any rate one rarely meets native British speakers in service industry jobs. My bias comes mostly from British sitcoms and comedians vs. American television and British professionals I deal with vs. American politicians and professionals. This may be a bad proxy, but the level of vocabulary used by the (mostly British) sports journalists on James Richardson’s football podcast is head and shoulders above what you find in American sports journalism, and football is not the sport of the elites in the UK. Americans are generally encouraged not to show their erudition in public.

    • The Nybbler says:

      Definitely. Some time ago I worked on a project for a Korean company, and we had a bunch of Korean-Americans (2nd gen, 100% ethnic Korean) on our team. Very easy to tell the difference.

      • AdamM says:

        You mean easy to tell the difference from Koreans who grew up in Korea?

        • The Nybbler says:

          Yes, very easy to tell the difference between the Koreans and the Korean-Americans. I see I’ve worded this badly; I was not working for the Korean company, I was working on a project for an American company where the Korean company was the customer. The Korean-Americans worked for the American company.

      • Sfoil says:

        I remember standing in line behind a woman in Korea a few years ago and realizing she must not be a local because of her hair.

        But yeah, the difference really is stark given that it’s completely non-genetic. Korean-koreans dress completely differently in every situation, have different body language, and use different (and more) skin care/cosmetic products than their overseas brethren. And of course the incredible prevalence of eyelid surgery on the women.

    • Null42 says:

      Russia and Israel strike me as both having fairly ‘macho’ cultures from what little I’ve seen (though this is my extremely limited experience as an American, and Russians and Israelis are welcome to correct me), so maybe there’s a swagger to the guys at least that stands out?

  6. JRM says:

    First of two: Hey, let’s fight data with anecdotes!

    The body cam study cited does not match my extensive discussions with cops and civilians.

    First of all, the primary advantage of police body cams is as an investigative tool. Juries can see what happened; we have exact quotes and exact replays of police-citizen interaction. This is very helpful in all kinds of situations.

    Secondly, the methodology isn’t what actually happens. Sure, it seems fun to have a shift of cameras and a shift of no cameras, and then you can compare the results. But this doesn’t actually optimize the results:

    With some cameras:

    Citizen: Hey, Officer Friendly called me a filthy Norseman! And I’m Swedish!
    Cops: OK, fill out this form.

    With all cameras:

    Citizen: Hey, Officer Friendly called me a filthy Norseman! And I’m Swedish!
    Cops: OK, we have the transaction on tape. Would you like to look at the video?
    Citizen: I have an important meeting.

    Also, time matters:

    Day 1:

    Helpful Citizen: Hey, cop, I smell bacon!
    Cop: I smell poo.
    HC: Your mother was a hamster!
    Cop: Your father stank of elderberries!

    After this is played about once in court, Cop tells her friends.

    Later:

    Helpful Citizen: I disparage your appearance and ancestors!
    Cop: That’s nice. Is there anything I can help you with?
    HC: I find you regrettable and unappealing!
    Cop: Great. Have a lovely day.

    And, man, I hear how much my argument sounds like a defense of power pose or something (it works, just not under these conditions… under these other conditions, I swear!) But I’ve followed an early adopter of bodycams and the complaint rate plunged. I’ve spoken to cops who say they act better.

    And even if they didn’t work *for this purpose* I would love them as an investigative addition. But I am fairly confident that they do. Problems with the study next.

    • yodelyak says:

      Right. NPR covered this particular DC study, here. And the headline makes you think, because it explicitly says, “no effect on police use of force or citizen complaints.”

      Well, but then you read the article and find the police force was already really closely monitored and (compared to the average police force in the county) also the police force was quite well-trained to respond with appropriate force only.

      So, the cams weren’t expected to make a big change. Also, this:

      “In his view, the cameras have helped his department enormously after contentious encounters like a recent one on Christmas, when police officers fatally shot a man who was brandishing a knife. Some had suggested the man was not armed, but Newsham says the video shows otherwise.”

      Also, the study apparently looked only at whether officers who had cams were involved in fewer excessive-force incidents than officers who didn’t. My expectation is that top-down level factors matter a lot, so once a lot of officers have cams (even if it isn’t all of them) the management will rely on training that aims to make sure they don’t generate too much embarrassing footage… and so there’ll be fewer excessive force problems across the board, which this experimental design wouldn’t detect. So you need to track before-and-after for the whole force, not just the individual officer performance with and without cams.

      Also, there’s this: “And while cameras have had neutral effects in Washington, D.C., Yu says the devices might have harmful effects in places with policies that allow officers to review footage before writing their initial reports of violent incidents.” That feels like a no-brainer. Why does any place allow police officers to control whether or not their own cams are on, or to review the footage from their cam before they write their report? The whole point was to force them to be honest, not to give them another tool for constructing narratives or making excessive force or other unreasonable behavior look reasonable.

      • jasonbayz says:

        “Well, but then you read the article and find the police force was already really closely monitored and (compared to the average police force in the county) also the police force was quite well-trained to respond with appropriate force only.

        So, the cams weren’t expected to make a big change.”

        Who exactly didn’t expect the change? “They didn’t expect them to make a big change,” and “after finding they made a big change, they decided that due to the monitoring and training they shouldn’t have expected a big change” mean different things.

      • fnord says:

        Link to the study itself.

        I don’t think explanations focusing solely on behavior, like training received by officers in both groups, are sufficient.

        The study tracks a number of things that could be expected to be changed by the evidence provided by body cameras, even in the absence of behavioral differences between the groups, such as “complaints disaggregated by disposition (sustained, not sustained, or insufficient facts to resolve)” and “whether MPD arrest charges are prosecuted by the U.S. Attorney’s Office (USAO) or the Office of the Attorney General (OAG) and the disposition of those charges”.

        My guess is that the body cams weren’t turned on (or the footage wasn’t kept) for critical incidents often enough to make a difference. The study found that “treatment officers uploaded about 665 videos annually” and “The average video recorded by
        a treatment officer was over 11 minutes long”. Assuming that the average video recorded by a treatment officer was no more than 12 minutes long (else, presumably, that would have been reported as “over 12 minutes long”), I calculate that as about half an hour of footage per working day. That leave quite a lot of time unrecorded.

        Still, that’s only a guess. My next leading hypothesis is that (as skybrian suggests) body camera footage mostly isn’t useful evidence, and there are probably further possible explanations.

    • yodelyak says:

      I think it’s easy to lose our brains with “well, but this study was a randomized trial” when what’s at stake here are things we should have very strong priors about. Video cameras *work* for reliably recording video footage, p < .0001. That's why high-speed cameras are useful for physics experiments; they’re quite reliable. If we believe that there is a police force in the country with a culture that creates serious excessive force problems, do we really think that letting the ACLU install 24-7 always-on cameras on each police uniform and cruiser wouldn’t likely do something to curb that excessive force problem?

      Baltimore has also been conducting a study on the efficacy of body cameras. Here’s how the study worked. They installed cameras that have an “on/off” switch, but didn’t tell the officers that the cameras are actually always on, and what the “on” button does is cause the video footage to start getting written to a permanent file–including 30 seconds or so of footage from before the officer pushes the button… which, if the officer is planting evidence, might catch the officer in the act. Sure, it’s not randomized. It wasn’t even intended as a study! But daaang, it is a compelling result!

      • skybrian says:

        Unaimed video cameras attached to moving bodies (such as a police officer who is running or in a fight) will tend to jitter a lot making results very hard to interpret and maybe even unusable. Video taken by a bystander with a mobile phone will often be much clearer and have a better perspective.

        The NY Times has a good interactive article that will give you an idea of the limitations.

        But I would guess some body cameras are better than others at stabilization? I don’t know much more about them.

        More generally, it seems like a lot of times we talk about technology in a political context without really learning much about it. It’s easy to just say “body cameras” and forget about real-world limitations and assume that technology works better than it actually does. It’s much like the difference between the science fiction and real-world version of a technology.

      • albatross11 says:

        On the other side, it’s important not to discard evidence that contradicts our priors.

        One thing that’s important to remember is that US policing is intensely local–almost all the police you ever deal with are employees of a city or county government, and different cities can have *massively* different cultures and histories and levels of competence and honesty. That makes it hard to draw a lot of conclusions from a study on one or two cities. But that’s true in all directions, not just when the study contradicts your prior beliefs. To the extent we’re concerned with differences in local police forces, we should be just as skeptical of studies that show positive impacts from body cameras.

  7. c0rw1n says:

    The narrative of bonobos as the sex-loving, peaceful, polyamorous alternative primate doesn’t stand up to scrutiny.

    … except when they’re all young and well-fed :

    … early research frequently focused on captive bonobos and artificial feeding sites. These groups were often quite small, had many sub-adult bonobos, and, of course, the captive bonobos were not living in natural settings. Juvenile and adolescent bonobos turned out to be far more sexually inclined than adults are. Rich concentrated food stores (artificial feeding sites) induce near-panic in bonobo groups and this provokes sexual behavior. It is not representative of typical bonobo life in the African wild.

    #Neoteny #Abundance #Domestication

    • cactus head says:

      >Rich concentrated food stores (artificial feeding sites) induce near-panic in bonobo groups and this provokes sexual behavior.

      Why do these things (near panic, sexual behaviour) happen?

      • Deiseach says:

        Pulling this out of my ear – a rich food source is very desirable, so the bonobo troupe (a) does not know this is artificial and will be maintained (b) expect the kind of thing that would happen in the wild – competitors (other bonobo troupes, other primates, predators and so on) will show up to contest for it.

        This is going to be stressful – think of somebody convinced he is likely to be burgled because all the neighbours know he has a huge pile of cash stuffed under the mattress – and that induces panic.

        Why panic and stress then induce sexual behaviour – well, no idea, but isn’t there an idea about people who have had near-death experiences/war time situations engage in a lot of sex (and other risky behaviours) because they feel like they’re living on the edge, live is short, grab what you can and to hell with convention?

        • herbert herberson says:

          My (very layperson) understanding is that bonobos use sex to restore/strengthen social bonds, so it makes sense that sexual activity would follow a period of group panic.

          • Aapje says:

            @herbert herberson

            It’s actually more a fight for status and resources, where bonobo’s use both violence and sex to improve their status and to get access to food or other resources.

            The ‘social bonds’ narrative that paints bonobos as some kind of weed-smoking, free love hippies is less accurate than a sleeping with the boss narrative.

          • herbert herberson says:

            Technically, you’re not really disputing the phenomenon of social bonds being strengthened so much as you’re doing a Marxist critique of the nature of the social bonds

          • Art Vandelay says:

            What would be Marxist about that critique? Marx is all about bonds and working together as a collective. It’s a more mainstream economics critique – this is actually about self-interest and trying to get ahead as an individual.

        • Aapje says:

          The dominant male apes gain control over the food source and then low ranking females trade sex for access to the food:

          Feeding is highly segmented by rank. Low ranking individuals may be charged or attacked if attempting to line-jump.

          Female bonobos disperse to other groups around adolescence. When accepted in their new group, they solicit sex acts from higher ranking males at food sites. This behavior, and their total libido, drops substantially as they gain rank. In order words, they’re forced to barter sex for food. Not because they’re eager for sexual contact. When they don’t have to, they don’t. Few consider coerced prostitution a sign of gender equality.

          https://www.skepticink.com/incredulous/2017/10/09/bonobo-myth-demolished/

          Anyway, the entire thing is worth a read and raises some very uncomfortable questions of how much of the ‘patriarchy’ is actually nature or nurture so closely linked to nature that it emerges unless we fight against it; rather than the common ‘fall of man/men’ myth where mankind started out completely egalitarian and then got corrupted into adopting patriarchy.

    • Ghatanathoah says:

      Unfortunately I don’t have the upper body strength needed to throw people very far. So I don’t think I should trust the research in Lynn Saxon’s book.

      Maybe I shouldn’t judge a book by its cover. But I can’t help but notice that the book cover looks like a cheap advertisement from the Seventies. Weren’t there any public domain photos of bonobos she could have put on it? And there’s the fact that the publisher listed is “CreateSpace Independent Publishing Platform.” And that “Lynn Saxon” is a pseudonym and no one know her real name. Those things don’t automatically discredit her work, but they do raise more red flags than the Soviet Union.

      Maybe I shouldn’t discount Saxon just because she’s an “independent researcher” who self-published her work and has no talent for cover design. But little signs like this are how you determine whether a book is worth your time to even read.

      • Edward Clint says:

        There are no “red flags”. Lynn Saxon is her real name, not a pseudonym. The content of the book, not the prettiness of the cover should be what matters. Also, the book has been endorsed by USC primatologist Craig Stanford (see his tweet in the article). Also, in a previous posting, I printed a letter from a primatologist who studies bonobos in the field who said, in summary terms, exactly the same thing (as have several other of my primatologist colleagues). Please do not spread these untruths further and if you disagree with the claims, then cite research… not fonts.

        • Jiro says:

          Being self-published, however, genuinely is a red flag.

          • I don’t think I agree. You might consider that all of Scott’s work is self-published, online rather than in print. Is that a red flag?

            I have four self-published books (counting two written jointly with my wife), one of which is the third edition of a book the first two editions of which were commercially published. Also six commercially published books.

            At this point, self-publication is no longer vanity press, it’s an inexpensive way of avoiding the hassles of commercial publication. Of the two books I am currently trying to get published, one will probably end up as a web page instead, for somewhat special reasons, and the other I will self-publish if none of the academic publishers is interested in doing it.

            It’s true that commercial publication provides some positive evidence that self-publication doesn’t, but it’s evidence that someone expects the book to sell a significant number of copies, not that what it says is true. For that you need to read it, or at least reviews, and make your own judgement.

          • Jiro says:

            You might consider that all of Scott’s work is self-published, online rather than in print. Is that a red flag?

            Sure. a red flag doesn’t mean “bad 100% of the time”. Also, a lot of the value of Scott’s work comes from people’s responses to them. Neither a self-published nor commercially published book would have that, not like we have here.

            At this point, self-publication is no longer only vanity press, it’s also an inexpensive way of avoiding the hassles of commercial publication.

            Fixed that for you.

            Self-publication is not always done for vanity press style reasons. But it is sometimes done for such reasons, and self-publication is Bayseian evidence that it was done for such reasons (even if that is not true 100% of the time).

    • Alethenous says:

      So it’s a bit like locking a bunch of human teenagees without supervision in a room full of sugar?

  8. JRM says:

    From the body cam study:

    For example, we find that a group of 1,000 officers with BWCs is estimated to report 74 more uses of force in a year than officers without BWCs. This is our best estimate. However, the data are also consistent with the real effect of BWCs being anywhere from a decrease of 97 uses of force to an increase of 244 uses of force per 1000 officers, per year.

    Oh, for pity’s sake. This sort of thing makes me think the authors are unfamiliar with the problems in DC, except that they are not because they talk about them. They really appeared to try, but not here.

    The DC cops have had all kinds of problems with improper use of force. (Baltimore’s got some issues, too. New Orleans is bad, but has improved some.) Now, you find that SELF-REPORTED uses of force go up with bodycams? Therefore, there could be increased uses of force?

    How about this alternative explanation: In DC, cops don’t always report uses of force when they don’t have the cameras on. It takes just a small percentage of cops with a lot of unreported uses of force to skew that data. The fact they don’t seem to realize this is a possibility makes me think the editing process was broken.

    I’ve now read the charts on results – they show the results on court cases to be positive, but not to a statistically significant degree. More study might find they are positive. I absolutely believe it would take a large sample size to show the effect – most cases are pled early and often with or without cameras. But there is a positive effect at the edges. And it’s at the edges that guilty people may go free, or innocent people convicted.

    One more anecdote: I had a case with a dude who was charged with bringing drugs into jail. The jailers asked him if he had any drugs:

    Citizen (very high): No! I never use drugs! Drugs are bad!

    Cops find some weed in his pocket as he’s getting processed in, and that’s quite illegal under California law.

    Turns out, he got his pocket tossed by another cop in the street who did an insufficient job; he probably reasonably believed he did not have drugs on his person on the way into jail. I found that out because the defense attorney called me and pointed me to the exact point on the bodycam footage it occurred. Case dismissed.

    Fold that in to increased citizen confidence, a gradual lowering in costs as an economy of scale grows and storage space gets cheaper… bodycams are good police procedure, and I think it is not a close question.

    And, yes, I still feel a little like I’m arguing feelings over science. But I think a complete utilitarian analysis and getting the data and its meaning right leads to my conclusion.

    But that’s just my opinion. I could be wrong.

  9. fortybot says:

    > If you like your questionable science served with a side of other questionable science, here’s MBTI Facial Phenotypes, the average face appearance for each of the Myers-Briggs types (eg “ESTJ”, “INFP”, etc)

    Hm, I wonder what carries over. Presumable some level of personality is transmitted through the images (because, for example, some are smiling, and some are not), but I’d be interested in how much anyone could glean.

    • Null42 says:

      What’s funny to me is you can see the ghosts of eyeglasses on the female INTP, but not the male INTP. Similarly, the male ENTP seems to have glasses, but the female ENTP doesn’t. Of course it reflects who’s going to invest in contacts to avoid looking too intellectual or nerdy, but it’s a funny derived characteristic.

  10. Sniffnoy says:

    I have to disrecommend the paperclip game. Aside from how long it takes, so many important things are not at all clear, making a nonobvious mistake can set you back hours, and sometimes it can basically just make the game unwinnable. I eventually gave up when it became clear that there was simply no way I could beat the number of drifters I had made, and that any attempt to improve my strength against them would just worsen matters.

    (Also, I feel like it gets the theme wrong in some important ways.)

    • skybrian says:

      I thought it was great, but it didn’t occur to me to try to finish it. It seems like this sort of game is more about the concept than about actually finishing.

    • reasoned argumentation says:

      Disagree – game is solid.

      2.5 hours to paperclip the Earth – which did get tiresome near the end when I was just waiting for it to finish making enough clips to get the last few points of trust / make enough money to get more trust.

      The “analyze the payout matrix to determine the best strategy” minigame was fun.

      EDIT: Apparently “full autonomy attained” wasn’t the end.

    • WashedOut says:

      Great game. I don’t know enough about economics to have a good intuition for the demand/price relationship, so I struggled to make ends meet earlier on but business was booming later.

      What determines the price of Wire? Looks like it takes a random value between $13 and $27.

    • Peffern says:

      I can’t bring myself to play cookie-clicker style games when I could be working on my Factorio complex

    • The Nybbler says:

      A few tips

      1) You need 70 memory (slightly less with quantum) and 100 total trust to get past stage one. So don’t buy too many processors.

      2) Total cheat: there’s no need to buy any strategies for the strategy game. They make it slower and more expensive. (I think in earlier iterations you could still get bonuses without all the strategies, but you can’t now. You can still get plenty of Yomi, because of 4)

      3) Total cheat 2: One photonic chip is quite sufficient.

      4) You can win without honor or any honor-related achievement. 20 total trust on the probes is plenty.

      • Nabil ad Dajjal says:

        Total cheat: there’s no need to buy any strategies for the strategy game. They make it slower and more expensive. (I think in earlier iterations you could still get bonuses without all the strategies, but you can’t now. You can still get plenty of Yomi, because of 4)

        This one really pissed me off when I figured it out myself.

        The time for each tournament is the square of how many strategies there are but the payout looks like it only grows linearly. So you’re actually losing Yomi/t with every additional strategy you buy!

        It would be much more reasonable if the payout grew with e.g. the cube of the number of strategies so that there’s actually a reason to buy those upgrades.

    • Rosemary7391 says:

      I’m kind of stubborn, and left my computer running overnight. I eventually managed to get enough trust to stabilise at ~200 probes:8 million drifters, then very slowly increased that… I “won” eventually. Not sure whether there is an actual end.

      • Edward Scizorhands says:

        There is an end.

        There are obvious UI improvements that could be made, but part of me wonders whether that is part of the game.

        For example, at the stage you are at, production moves at the slowest of the bottleneck of 1) discovery 2) acquisition 3) wire-making and 4) factories. But you can’t see the total amount of capacity you have of each stage; you can only see what the current output is. And the stages aren’t in order on the UI, nor are those build steps in order on the production list.

        Maybe part of the game is figuring all that out. It can be quite cruel to exploring options, though: I tried to gauge how fast I would lose ships with various drone settings, and I turned hazard mitigation down to 0, and lost the whole fleet in a matter of seconds.

        Does it matter if you are outnumbered by the enemy? I was able to scrabble up from a serious disadvantage to maintaining a 100:1 advantage, but maybe it makes no difference at all.

        • The Nybbler says:

          Does it matter if you are outnumbered by the enemy? I was able to scrabble up from a serious disadvantage to maintaining a 100:1 advantage, but maybe it makes no difference at all.

          The ratio of ships in a particular battle is related to the ratio of ships total, and having more ships in a battle gives you an advantage.

          • moonfirestorm says:

            It doesn’t seem to map perfectly though, and there’s some random factor: despite having something like 100:1 advantage in drones, I don’t think I ever saw even a 10:1 advantage on the battle screen, and every now and then I’d run into a battle where I was pretty significantly outnumbered.

            Somewhat frustrating when the only way to get decent amounts of honor is to have a massive winning streak, and guaranteeing that needs SERIOUS focus on combat. I assumed it was going to be like the tech in stage 2 where you get a production multiplier that resets if you ever go under power: that is, an incentive to maintain decent combat stats because you’d sacrifice a large stacking bonus if you neglected it even for a short amount of time.

          • Witness says:

            Once you have the OODA loop, Speed is also a relevant combat stat. It’s still annoyingly possible to get massively outnumbered and lose a fight, but splitting some of your combat points off into speed can improve your odds pretty dramatically.

          • Brad says:

            You can get all the honor you need with yomi and the songs. Your max trust doesn’t need to be all that high to win.

          • moonfirestorm says:

            Yeah I mean I got through the phase in pretty short order anyway: running something like speed 5, combat 7 won me the vast majority of battles I ran into, and the speed helped tiling the universe fast too.

            It’s more that they felt like undeveloped systems. I think I ended up with like 2 honor upgrades by endgame, and was never able to go above 0-1 in a lot of the probe stats because Hazard/Speed/Combat/Replication sucked up so much of my design points. I was pretty excited to improve my probe designs eventually, but I ran out of universe before I got to upgrade my drones to really do anything other than survive and win the war.

            Maybe that’s the point, and I sorta liked the paradigm shift where paperclips matter so much less at endgame, but I would have liked them to matter at least a LITTLE bit (I don’t think my unused paperclip total ever stopped going up through phase 3). After like 5 minutes with the 3 paperclip supply parameters, those basically stayed at 0 for the rest of the game.

          • Witness says:

            I don’t think there’s any way for self-replication (+ defenses against probe loss, including by combat) not to dominate for most of this stage. Exponential growth is hard to beat, so most of the time your upgrades go there.

            I guess if they made the replication more expensive, you would need to spend more time with factories/harvesters/wire and maybe exploration ticked up, but this I think would have been more annoying, not less.

          • The Nybbler says:

            You can win without honor. In fact, I’d go so far as to say honor is a trap; it lets you increase probe trust, but increasing probe trust increases drifter creation rate.

        • Rosemary7391 says:

          Oh, no. I was going to quit after this round! Now I feel like I want to find the actual end…

          I didn’t mind the UI much; it was kind of fun finding all the different options at least the first time through.

          It does matter if you’re massively outnumbered. I had to get probes to replicate quickly enough that they literally couldn’t be killed that fast. It took a very long time!

          • Edward Scizorhands says:

            Is there any way to lose? I played a quick game last night trying to lose as fast as I can, and it kept on throwing mulligans at me to make up for my stupidity.

          • Witness says:

            @Edward

            I do not think there is a way to lose.

          • Doctor Mist says:

            I do not think there is a way to lose.

            The first time I played it, I thought I must have been close to the end because I saw it offer the project of sending out probes to colonize the universe and figured that was the big finish. (How little I knew.) But I didn’t come even close to meeting the requirement of 5 octillion clips, so I gave up and closed the window. In retrospect, it might just be that it didn’t occur to me to start disassembling my infrastructure, and maybe that’s always enough to put you over the top. So you can lose, in the sense that you can be too stupid to win.

            Does it count as losing if you accept the deal at the end? I didn’t, and found the outcome oddly satisfying.

    • moonfirestorm says:

      I think it kind of steers you towards making the right choices, though: throughout at least stage 1, you generally have a couple upgrades that need way more ops than you have (with hypnodrones showing up fairly early and dangling above you at 70k ops). That should steer you towards memory pretty consistently, although trust becomes hard to find later on and it’s possible it doesn’t steer you far enough. It’d be nice to be able to swap processors with memory and back, although I wonder if that would lead to some weird behavior where you constantly trade your memory out to fill your lower ops faster, then swap back to memory. 30s cd or something on the trade would cover it without locking you out of the game for more than like 10 minutes.

      I loved the game because mentally it feels very similar to one of my favorite games, Factorio. It’s very much a game about rates rather than quantities, and spotting the bottleneck in your progress chain, which will change naturally as the game progresses (I was expecting Stage 3 to be a lot more about paperclips than it was).

      • Witness says:

        If you build up enough creativity, there’s a project that lets you redistribute your procs/memory. I don’t recommend going for it, it seems to just be there in case you get “stuck” in stage 1.

        • Standing in the Shadows says:

          That option appeared for me today this morning, and got me unstuck. I had the hypnodrones out in short order after that, and am now carefully ramping up power plants, matter gatherers, wire makers, and clip factories.

        • moonfirestorm says:

          Kind of a neat way of doing it too, because the typical screw-up is too many processors and not enough memory, and that’ll inevitably result in a lot of accumulated creativity while you try to build trust through paperclip number to get yourself out of the hole.

    • Error says:

      Just how many drifters did you have?

      On my second playthrough I deliberately ramped them up to about 10e12 and then let them kill all my probes, jut to see if it was possible to get “stuck” on the combat mechanic. I came back just fine. It took a bit of twiddling to find a balance of replication/hazard resistance/speed/combat that kept my numbers going up despite the battles, but it was quite possible.

      Given that I came back with just ~100 new launches vs a quadrillion drifters, I assume the game formulas allow for the same vs. any arbitrary enemy count.

  11. philkidd says:

    The EY post on AlphaGo Zero has a number of problems. The claim quoted above, “the mighty human edifice of Go knowledge…was entirely discarded…” is highly misleading. In fact, given that it started without training data, it is remarkable how human-like AlphaGo’s game is. It rediscovered a number of known openings, and the new openings it plays are very similar to known openings played by modern players. Some of them are actually old openings that modern pros had discarded as suboptimal (presumably erroneously).

    EY also emphasizes that it took “centuries” for humans to master go. There is no historical data to support this claim. Certainly, opening theory and style of play have evolved over the ~350 year recorded history of go. However, the difference in strength between modern pros and the best players at the beginning of recorded go history (e.g. Dosaku in the mid-17th century) is relatively small, and, as I mentioned above, AlphaGo may lead to a re-evaluation of some older moves that were previously thought to be bad.

    Finally, EY and Robin Hanson both fail to appreciate the degree to which AlphaGo Zero’s architecture is specialized for playing go, even though it did not include go game data. The training procedure for AlphaGo relied on a search method called Monte Carlo Tree Search (MCTS). MCTS works by simulating large numbers of games with random moves, or in the case of AlphaGo, random moves chosen using neural network evaluations as a weight. This strategy works particularly well for go because stones do not move once placed, so a randomly chosen good move can influence the state of a game for a long time, even if it is followed by many bad moves in other places. This is not the case in chess, for example, where MCTS does not work well, because a randomly chosen good move is likely to be undone by subsequent bad moves in a random playout. If the AlphaGo architecture does not even generalize well to chess, it would be premature to call it a real step on the road to AGI.

    • BlindKungFuMaster says:

      “However, the difference in strength between modern pros and the best players at the beginning of recorded go history (e.g. Dosaku in the mid-17th century) is relatively small …”

      Really? The difference between chess players from just a hundred years ago and today is massive.

      • philkidd says:

        What do you mean by massive? For sure someone like Capablanca would be tangibly weaker than Kasparov or Carlsen, but this would be almost entirely due to differences in opening preparation, and Capablanca would (I think) still be one of the strongest players in the world, if not quite at the very top. The situation in go is comparable, except that opening preparation is much, much less important in go, so Dosaku would be (again, I think) quite close in strength to a top modern pro. Also, opening theory in go didn’t really start to evolve much until the mid-20th century, 250+ years after Dosaku, driven by changes to the rules about handicaps and scoring.

        • fredtwilight says:

          I’m going to have to respectfully disagree about the chess one.

          Sure the easiest difference to measure these days would be the level of opening preparation however, that is to some degree a reflection of our greater understanding of the game in general. People are aiming for different types of positions, and contributing to the massive body of theory simply because people have to learned to defend certain kinds of structures in a way they never succeeded at classically. (Helped ironically, very heavily by the computer).

          Sure you give Capa a couple of years to adjust to this and he would be incredibly strong. However this isn’t something that he would think up on his own if it wasn’t being implicitly shown in the way we play the game now.

    • actinide meta says:

      His description of Zero’s architecture is also wrong in an important way. Zero is NOT a neural network that selects moves; if you use its policy network that way it is much weaker [1]. It is (still) an MCTS driven by a neural network to suggest the prior probability of moves (“policy”) and an estimate of the outcome of a game state (“value”).

      I think the class of architectures that Zero belongs to probably is important, mostly because they are the first RL applications of neural networks that seems to be at all stable in training. But clearly zero itself is not a very general agent, and would probably struggle to learn to play any game that doesn’t have a lot in common with go. I think it’s fair to claim that they didn’t teach it a lot about human strategies for playing go, but they definitely built it (with lots of trial and error!) to play that particular game.

      The most plausible scenario I can think of for “foom” involves architecture search turning out to be a “narrow” domain like Go rather than a “AGI-complete” domain like programming in general. Then someone finds the right architecture for architecture search, and like Zero it blows past human ability in the narrow domain of optimizing RL agents. Then *it* quickly makes some breakthrough(s) to create superhuman AGI. Call it the “non-recursive self improvement” scenario.

      [1] The Zero paper actually gives performance measurements for greedy play based on the policy network. It is weaker than AlphaGo Fan, so presumably weaker than top human players. Still impressive!

      • Ilya Shpitser says:

        Look, if folks worry about Zero, they should make specific claims they are worried about.

        You may notice that EY is very careful NOT to make any falsifiable claims he’s talking about. I think the reason for this is this is how he does fundraising. Rather than apply for an NSF grant, or something, he writes a blog post to create generalized worry and create a MIRI funding stream down the road that way.

    • Shion Arita says:

      The key (reason that alpha go poorly predicts how ‘real’ strong AI would be) is also that Go is a defined problem, where the map is the same as the territory. Explosive move tree growth aside, that’s the kind of thing that computers are really good at inherently. They’re not so good at less well-defined problems.

      I’ve long thought that the rate-limiting factor to progress is largely in getting data, and that’s why while AGI will be very helpful and competent it won’t explode into omnipotence like EY and others think.

      Sure, human knowledge is kind of crappy in a lot of ways, but it’s also hard to do better than it. Coming up with a good go move is easy. Designing an experiment that will distinguish between two theories out in the real world with all its messiness is much less so. Especially if those experiments take a long time to run. For example, tell me how the AI would be better at observing gravity waves.

    • David Speyer says:

      A question for those who know about Go: As far as I can read, Alpha Go Zero has only demonstrated its strength against other earlier versions of Alpha Go. Ordinarily, sports rankings are strongly transitive, so the fact that Alpha Go Zero trounced Alpha Go Lee 100:0 and Alpha Go Lee beat Lee Sedol 4:1 is evidence that Alpha Go Zero is vastly better than all humans. But we’ve never seen something like Alpha Go Zero before. How possible is it that Alpha Go Zero is simply incredibly optimized at beating opponents who use MCTS on neural nets?

      • Brad says:

        I’m not a go expert by any means but the paper mention that zero recovered on its own some strategies and structures developed by experts. So I don’t think it was like an adversarial image for a image recognizer that just exploited the underlying NNs without actually playing the game.

  12. Wrong Species says:

    On Google, I can’t believe it has such a higher rating than the others. All of them don’t really care about your privacy, but Google is the worst offender, trying to know every possible fact about you to use in their marketing. And the fact that they have such a wide access to peoples phones, search history, emails and even their entire web browsing makes it that much worse.

    • Scott Alexander says:

      Theory: nobody cares about privacy. Facebook and Twitter are annoying time sinks who keep messing up your timeline order. Google is the nice company that helps you find whatever you’re looking for.

      • Watchman says:

        Speaking as someone whose professional duties include reminding my department that data protection is a thing (with big fines attached to it), I think there is a hell of a lot of evidence to back you up there.

        If Google was selling information to criminals, then there might be an issue, but seemingly most people are not concerned about third party acquistion of personal information, especially if that transfer of information actually improve the general experience of the first party through more targetted information. Although as someone who keeps getting invitations to MBA briefings in Jakarta because I happen to be responsible for some relationships with Indonesia, I would suggest there is the potential for this relationship to go bad quite quickly…

    • Kaj Sotala says:

      And the fact that they have such a wide access to peoples phones, search history, emails and even their entire web browsing makes it that much worse.

      Introspecting on my emotional reactions, I think that on some level this actually makes me like Google more. They know a lot about me and use this knowledge to help me out, and also haven’t betrayed my trust so far… which makes my relationship to them feel intimate, in a way that Facebook doesn’t.

      I think people instinctively trust those who could hurt them but have instead only been helpful, which would go a long way to explaining Google’s popularity.

    • Whatever Happened To Anonymous says:

      Well, one thing to consider is if it’s X% favorable opinion vs unfavorable or vs unfavorable or indifferent, a lot more people don’t use Facebook and Amazon than Google. Facebook has also the added problem that it’s a “social network” so a lot of people don’t like it on principle, and that a lot of young people don’t like it because it’s full of old people (migrating to other networks whose grand selling point is not being Facebook).

    • quaelegit says:

      Agree that Scott and Katja’s ideas seem plausible. People might also be misinformed. I thought Fb was much more invasive about data commercialization than Google.

  13. Markus Karner says:

    Let me cast some doubt on the bonobo “debunking”. This seems to be the work of a serial debunker, not of high quality, and definitely not science in any recognizable way. Cursory review of the reviews of her books on Amazon suggest she’s got an axe to grind with anyone making light of biblical monogamy etc. Her other book “debunks” “Sex at Dawn”, a book that talks about humansexuality, and apparently critiques it in similar spiteful fashion. So we’re talking politicized polemics here.

    I happen to have read both “Sex at Dawn” and the more scholarly work on Bonobos, both the objects of scorn of this debunker, and I don’t recognize the works from the criticisms. Frans de Waal’s work on bonobos in particular is quite sober on the ins and outs of bonobo behavior, even describes a political attack and murder of an opponent for group leadership – no sugarcoating here. So, the criticisms of this debunker sound shrill and like a lot of strawmanning. Handle with care.

    Didn’t you do a post “Against debunking” at some point, or similar? Came to mind on this subject.

    • Scott Alexander says:

      Thanks; removed until I can confirm.

      • Edward Clint says:

        Lynn Saxon is indeed a debunker, which is precisely what is needed when distortion of truth are spread by the media.

        As a UCLA bioanthropology grad student, I taught primatology and have personally spoken with several primatologists on this issue. In an earlier blogpost, I published a letter from a primatologist who studies bonobos in the field (see https://www.skepticink.com/incredulous/2014/12/29/questioning-sexy-bonobo-hype-part-2-primatologist-responds-christopher-ryan/) who says exactly the same thing. Saxon’s book was also endorsed by USC primatologist Craig Stanford who called it a “thorough review” (see https://twitter.com/craigstanford7/status/719739088123748354) as well he should, because Saxon meticulously supports her points with published, peer-reviewed bonobo research.

        “Sex at Dawn” is pop science drivel, but hey, why listen to published academic research when you can take your lead from a guy with a distance learning PhD from a regionally-accredited program who has never published academic research of any kind.

        • Markus Karner says:

          This makes me chuckle a little now. You’re right (your other comment), I haven’t read Saxon’s books, that’s why I didn’t say “Saxon is wrong” but “I’d be careful here”. And, I read Ryan’s and several of de Waal’s plus assorted works on bonobos etc, I’m originally from biology too.

          So here it goes – the way different writings come across, that’s what makes me chuckle. I read Ryan’s book with a grain of salt because it’s a pop book compiling science in a journalistic fashion. So I applied journalistic standards of “truth” and found it interesting, but I’d never call it “real science” and so I wasn’t offended, just felt stimulated in some ways, disappointed in others, and that was that. Basically, even the attempt at debunking it shows a lack of sense of humour and putting things in the right perspective.

          Frans de Waal’s books on the other hand often offer detailed and careful descriptions of observations and a lot of what i would call science. And then, you make of it what you want. Whatever he might imply about society somewhere else I don’t know but I was happy to see what I read as descriptive and scientific, rather than normative and polemic.

          Your endorsing review, “Faux-nobo: ‘Naked Bonobo’ demolishes myth of sexy, egalitarian bonobos” however, all the way from the title down, rubbed me the wrong way as sensationalist, eyeball grabbing, and triumphalist – moralizing (about someone else’s “bad science”, not about societal mores). Journalism again, but the aggressive, proselytizing kind. (unlike Sex at Dawn).

          I keep an open mind, maybe I’m wrong here and the debunking was justified. I really dislike the romanticising of science. But I equally dislike the gushing pleasure of popping someone else’s dreams (not mine, here) that many debunkers appear to show, a kind of sadistic “your belief is rubbish” Schadenfreude that seems to be their primary motivation, not some commitment to “truth” (sorry, I’m not romanticising “truth” either, there are many dimensions of insight). Sorry if that wasn’t your intent but this is how “Faux-nobo” came across to me – smug and sensationalist.

          • Edward Clint says:

            >I didn’t say “Saxon is wrong” but “I’d be careful here”.

            You went quite a bit further. Let me copy and paste for you.

            This seems to be…not of high quality, and definitely not science in any recognizable way.

            This doesn’t seem like a reasonable criticism prior to actually reading a book.

            Your endorsing review …however, all the way from the title down, rubbed me the wrong way as sensationalist, eyeball grabbing, and triumphalist – moralizing

            Except that you don’t know that it is sensationalist. Sensationalising means to be provocative at the expense of accuracy. But you haven’t read the book. You don’t know that what I’ve written isn’t accurate, by your own admission. You think it is, and go from suspicion to dismissal without delay.

            I will plead guilty to the charges of grabbing eyeballs and “triumphalism”- but I call them being engaging (this is a blog post… not a dry academic pub. It’s meant to stimulate discussion… which it did), and passion. But Saxon deserves the accolades… anything less would have been inappropriate. Credit must be given where credit is due. You might not believe this, but I read through my draft post a couple times specifically to analyze if the language was overly strong. What I left after editing was what I felt was tenable and merited.

            Journalism again, but the aggressive, proselytizing kind.

            Well, let’s think about that. A good, “aggressive proselytizer” seeks to convert. The object is a political one, so the tools are rhetoric and persuasion. Such a person might engage in effective modes of rhetoric, like casting suspicion and doubt. Character assassination. Insinuation of ulterior motives. Someone quite different, who wants open discussion in the interests of science and reason will privilege evidence as an external referee, not rhetoric. They’d appeal to subject matter experts and make copious use of citations of legitimate scientific research. They’d even gladly agree to a cordial debate with a dissenter.

            But I equally dislike the gushing pleasure of popping someone else’s dreams

            I know just what you mean. And here, at least, we’re in total agreement. You know how I first got involved with all this? A few years ago, a friend of mine invited me to be on his podcast with Christopher Ryan. He liked Ryan and his book (it gelled with many of his beliefs and values). At the time, I knew that there was skullduggery afoot, but I did not know how deep it went. After that podcast, I started talking to my primatologist colleagues about bonobos. This lead to the first writing about it. I was shocked, frankly. And I was really upset because it meant that good friends of mine were being lied to and manipulated. My friends are good people, as I believe are others who have been taken in. I have nothing but sympathy for them. I am trying to help them.

        • Markus Karner says:

          Clint, seems like I can’t reply directly to post Nov 1 7:46pm so i do it here.
          I’d have to disagree with still quite a few things but no point to tit for tat this out endlessly. Just one item by way of example

          ‘This seems to be…not of high quality, and definitely not science in any recognizable way.’
          This doesn’t seem like a reasonable criticism prior to actually reading a book

          I conflated your endorsing review with what I could reasonably conclude the book would be like. In other words, tone and style of your review actually hurt the credibility of the book here. For the record, when I (later) found your previous Bonobo posts, I found them somewhat easier to swallow.
          Still scratching my head on why Ryan is such a central figure, it’s not like his thesis was all based on bonobo behavior.
          Finally

          I have nothing but sympathy for them. I am trying to help them.

          That’s exactly the kind of condescension that puts people like me off. Even though, saving grace, I now can see better where you’re coming from, why the bonobo stuff matters so much to you.

          • Edward Clint says:

            I guess there is a maximum thread depth.

            In other words, tone and style of your review actually hurt the credibility of the book here.

            I’d say it’s fair enough to dislike a given style. No writer or approach to writing is everyone’s cup of tea. But you would be prudent to consider that the messenger doesn’t determine the veracity of the message.

            Still scratching my head on why Ryan is such a central figure, it’s not like his thesis was all based on bonobo behavior.

            Because he has been very successful in peddling misinformation. He wrote a NYT best-selling book. His TED video has 2 million views. Dozens of large and small media sites have quoted and cited him. He’s the single-biggest spreader of bonobo misinformation (but he is far from alone in creating the problem).

            That’s exactly the kind of condescension that puts people like me off.

            I am confused about why you think sympathy for people who have been lied to is condescending. We can all be, and likely all have been, victims of the lies and fraud of others. I sure have been. This isn’t about being smart or educated or otherwise “better”. Newsweek and Dan Savage have endorsed SaD- otherwise reputable sources that I’d generally take to be reliable. Everyone can be had. And people who think that they can’t? They’re among the easiest to take.

    • Deiseach says:

      No particular axe to grind here, save that there really was a fad for describing bonobos as the perfect peaceful society (look at people calling themselves “bonobo rationalists”)- matriarchy! sex! no fighting! and it never sounded that great to me: compulsory sexuality in all interactions? Translate that over to human social life and see if it sounds so paradisaical to have to give handjobs, blowjobs, sex, frottage, etc to everyone and anyone you meet in the course of your day when interacting with them.

      I don’t know if the debunking is religiously motivated or what, but I think it’s no harm if the pop culture notion of “bonobos can teach humans how to live in harmony” gets whacked on the head with a blunt object, dismembered and stuffed into bin bags, and dumped in the canal. Maybe then we can reach a middle ground on “bonobo society can be just as violent, self-interested, status-seeking and gender-conforming as any other, and sorry, there is no perfect paradise of the animal kingdom to give us an example of what we should do”.

      • Ghatanathoah says:

        I think an actual middle ground would be “Bonobo society is just as flawed as other societies in most ways and shouldn’t be copied wholesale, but it might have a couple good ideas that are worth trying out.”

        • Watchman says:

          Surely we need to specify what those good ideas would be? It would be equally misleading to present an image of any primate society as offering a buffet of good ideas that you could select from to suit your taste as to holding that whole society up as a model for us to follow. And much more irresponsible, since you would be allowing any form of extremist to take the ideas that support thier view (Chimpanzees practice inter-tribal violence – lets copy them…).

          I’m yet to understand why anyone would take a resource-poor, labour-intensive lifestyle, be it Bonobo, hunter-gatherer or agrarian, and hold up these conditions as a model for our society, which has as an idealised model (which is what is being discussed here) actually conquered all the problems being dealt with anyway – it is only a non-ideal model of western society that sees patriarchy as a problem to be dealt with for example, or immigrants as an enemy to be fought.

      • Markus Karner says:

        Well it depends what you are trying to achieve, to pop a popular romanticising of some research or the research itself. What I see here is that on the surface, Saxon tries the former, but does it so aggressively and brutally that she just ends up strawmanning the actual research and comes out triumphantly slaughtering both. I just don’t see how this is a way to do scientific critique. This is just about creating partisan polemics.

        • Edward Clint says:

          Saxon makes no effort at all to “pop” the research itself. It is the research that she is presenting in The Naked Bonobo. I have no idea how citing research to relate what is known about bonobos can be called “slaughtering” it. I take it you have not read this book. It would be prudent to refrain from such editorializing about a book that you aren’t familiar with.

        • Can you offer examples? Where in the book does she misrepresent the work she is attacking?

    • Gilbert says:

      I’m gathering from some later comments that we’re talking about Lynn Saxon, the “other book” being Sex at Dusk, a debunking of Sex at Dawn.

      I read those a few years ago and have some comments:
      – I don’t actually know who Lynn Saxon is (it clearly is a pseudonym) nor who might know. And Sex at Dusk is self-published. Those are valid negative pieces of evidence.
      – As for her bias, monogamy yes, “biblical” absolutely no. Dusk is almost as annoyingly anti-Catholic as Dawn.
      – Some of that may be pre-emptive signalling, because Dawn actually has a narrative about hidebound official science and Catholicism combining to hold their heroic heretical truth down. So Saxon may be extra vocal about not liking Catholicism either.
      – Wether the anti-Catholicism is strategic or not, Saxon is clearly a liberal feminist science lady pissed of about science misrepresentation. My guess would be a (at the time) grad student in something related.
      – And as such she has every reason to be pissed off, because Dawn actually is a steaming pile of shite. Saxon actually proves basically every important claim misrepresents actual science, mostly by giving crucial details Dawn leaves out. In other words this is a successful debunking. These, too, do exist. Mostly if the debunked thing actually is clearly bunk.
      – Of course debunking is a genre and Dusk is clearly written in that genre rather than that of a neutral scientific evaluation. In other words Saxon does have an axe to grind. But that ax is not even remotely traditionalist, it’s mostly orthodox science with some admixtures of feminsm.
      – And several Amazon reviewers rounding that to a conservative Christian motivation is actually fairly symptomatic of Dawns brand of cookery. That is the designated out-group to which opponents must be matched.
      – I havn’t read The Naked Bonobo so I can’t say anything about that.

      (For the record, I am very Catholic and actually might match Dawns theory of its therory’s opponents.)

      • Markus Karner says:

        I haven’t read Dusk (the debunking for Dawn), but I have read Dawn. My reading is clearly different. I thought Dawn was a thoughtful series of questions along the lines of “are we getting this right?” and it really didn’t feel like a polemic to me. Quite the contrary, I found Dawn frustrating at the end because it doesn’t show a clear way forward from the collection of results it presents. It felt more like a conversation with pointed questions. I’d agree that it is a bit uneven, the way some books are, even good books. Yet – if you look at the reviewer statistics of Dusk on Amazon and the detailed takedowns, there’s more than a few reviewers taking offense at the method. Many there accuse Saxon of citing research that proves the original points of Dawn rather than refuting it. That, my original reading of Dawn, the vitriolic style of the bonobo takedown and my original reading … I mean I don’t understand the romanticising of the bonobo either, I never read any gooey-hippie style conclusions into the bonobo research, but precisely to the point, why not just point that out instead of spewing all that hatred. I’d take Saxon with more than a grain of salt.

      • Edward Clint says:

        Lynn Saxon is her real name, not a pseudonym.

        Unlike some of the other books you mention, The Naked Bonobo is a relatively restrained review of the field literature. It’s light on rhetoric, and heavy on statistics and field observations of bonobos.

      • I don’t actually know who Lynn Saxon is (it clearly is a pseudonym)

        If you don’t know who she is, how do you know her name isn’t Lynn Saxon?

        • Gilbert says:

          If you don’t know who she is, how do you know her name isn’t Lynn Saxon?

          You’re right, I didn’t really know that and having been corrected by Edward Clint I’m now convinced it isn’t true.

          I think I googled her back when I read the book and came up with nothing. Some of the Amazon reviews complain about her using pseudonym and since I don’t know who she is I seem to have automatically merged my lack of knowledge and that allegation into a “fact”. Which is really dumb, because it would have been the only fact in those reviews.

          And then I acknowledged that “fact” because of a misguided sense of fairness.

          Sorry for that.

  14. OptimalSolver says:

    Research in Journal of Personality and Social Psychology: People do better than chance at matching people’s faces with their names

    This was the journal that published Daryl Bem’s precognition paper, but refused to publish the several follow-up studies that failed to replicate his results, so I’ll assume this is yet another finding waiting to be consumed by the replication crisis.

  15. Antistotle says:

    The only time you’re allowed to sign up for Obamacare insurance policies this year is between November 1 and December 15 (slightly longer in some states). This is poorly advertised possibly as a sinister Republican plot. If you’re an American and you need health care, check out the Obamacare FAQ sometime during that period so you don’t get left behind.

    The open enrollment period has been a mis-feature of a awful implementation of a horrible idea from the very beginning.

    And it’s not just for plans purchased on the exchange, it’s for *all* private insurance plans. There might be some exceptions for employer plans, but every single ACA compliant health insurance plan is getting altered over the next two months.

    My insurance is going from ~850 a month to ~1040 a month for a plan that is also going from ~6800 per person deductible to ~7300 per person (that might not be entirely accurate. When I say the monthly charge I knew I could no longer afford insurance, and stopped paying close attention).

    Let me repeat that–As of 1 January I, my wife and my daughter will no longer have ACA compatible insurance. We cannot afford to spend over 12 percent of our take-home pay on something that provides no value until we spend somewhere between 13000 and almost 15000 dollars.

    As to it being Trump, let me quote from an email exchange from 2014 (well before Trump threw his hat in the ring BTW):

    Sorry for the email. I just want to keep you updated with the upcoming cut off dates for health insurance.
    Tomorrow is the cut off for a April 1 start date.
    On March 31st 2014, will be the last day you can purchase health insurance with a start date of May 1st.
    After that day you will not be able to buy health insurance unless you have a life changing event until 2015.
    This applies to on or off exchanges.
    If you would still like some help or advice on this please reach out to me.

    The ACA has been a train wreck from the word go.

    The Democrats wrote the ACA without any input, they hired cronies to build the website, Obama issues constitutional questionable orders to defer implementation of key parts of the bill, arranged extra-constitutional payments to insurers to keep insurers from leaving the program or going out of business. It hasn’t bent the cost curve *AT ALL*, but it’s made it difficult for many people to find doctors, and it’s forced people to *change* doctors.

    It was, and is a HORRIBLE bill. But hey, we forced a few more people onto a government entitlement program. YAH!.

    • suntzuanime says:

      To be fair, Trump was supposed to repeal and replace Obamacare, and he didn’t, so now all of its myriad failures can be legitimately attributed to him.

    • cassander says:

      the worst part is that it was totally unnecessary. a large majority of the increase in coverage from the ACA came from the medicare expansion. the expensive, Rube Goldberg assemblage of exchanges and their accouterments has had minimal benefit.

      • MrApophenia says:

        Yep. The actual effect of Obamacare since it went into effect has really been a fantastic argument for single payer.

        • cassander says:

          Central planning failed again is a good argument for more central planning? Or was it the inability of CMS to build a functioning website on time and budget?

          Regardless, single payer remains catastrophically un-affordable.

          • albatross11 says:

            Distinguish between:

            a. We can’t afford to provide everyone medical care.

            b. Single payer is an especially unaffordable way of providing everyone medical care.

            Perhaps (a) is true. But the Obamacare mess looks to me like an argument against (b), at least by demonstrating that an obvious alternative to single payer doesn’t seem to have worked out too well in practice.

            Now, I suspect the real problems with Obamacare are less the big-picture model (mandatory health insurance with subsidies for poor people and regulated minimum coverage works out okay in several other countries), and more that the implementation had to keep every interest group in the medical industry happy.

          • Conrad Honcho says:

            I think we’ll eventually determine the rule of healthcare is that it can be:

            1) Good

            2) Not catastrophically unaffordable

            3) Universal

            (Pick any 2)

          • herbert herberson says:

            Also, the part of Obamacare that was closest to single payer, the medicaid expansion, is the part that worked best (at least where state gov’ts let it)–it was the attempt to reach universal coverage through the private sector that caused all the headaches.

          • Iain says:

            How do you reconcile the “catastrophic un-affordability” of single-payer health care with the existence of numerous single-payer systems around the world that somehow manage to muddle through?

            The US spends more on health care
            (both per capita and, I think, as a percentage of GDP) than any other country. If US private spending and US public spending were two independent countries, they would both spend more than the OECD average. Many OECD countries have single-payer health care. To the best of my knowledge, all of them have something closer to single-payer than the US. (Certainly all the prominent ones.)

            Compare the Medicare expansion to the exchanges. Which of the two would you say was more centrally planned? Which would you say was more market-oriented? Which one worked better?

            Edit: Scooped by herbert herberson.

          • Paul Zrimsek says:

            Now, I suspect the real problems with Obamacare are less the big-picture model (mandatory health insurance with subsidies for poor people and regulated minimum coverage works out okay in several other countries), and more that the implementation had to keep every interest group in the medical industry happy.

            This is a pretty strong argument in favor of cassander’s (b), isn’t it? Under single-payer, all health-care spending will funnel through an institution which has to keep every interest group in the medical industry happy.

          • cassander says:

            @albatross11

            Define “medical care”. We could certainly provide everyone with all the vaccines and antibiotics their doctors will give them. We can’t send everyone to the world’s number one ophthalmologist every time they need a new prescription for their glasses.

            But the Obamacare mess looks to me like an argument against (b), at least by demonstrating that an obvious alternative to single payer doesn’t seem to have worked out too well in practice.

            One, I’d hardly call the the ACA an obvious solution to any problem. the ACA was a hodgepodge designed from the beginning to minimize political opposition, not achieve serious reform. The very fact that it had an employer mandate should more than prove this, I’m not aware of any serious healthcare economist, left or right, who defends employer provided care as anything other than a disaster. But including it made the ACA seem cheaper. You’re not wrong about concept vs. implementation, but you can’t just hand wave implementation, because the same implementation problems will exist for any alternatives.

            Two, the fact that the ACA failed doesn’t make single payer more affordable. Even proponents of single payer admit for it run to the trillions

            @Iain

            How do you reconcile the “catastrophic un-affordability” of single-payer health care with the existence of numerous single-payer systems around the world that somehow manage to muddle through?

            One, actual single payer systems are relatively rare. Most countries use some sort of multi-payer system.

            Two, we don’t have to guess what a US single payer would look like, it has one called medicare. We know very well how much it costs, and that it has not been able to restrain costs the way other countries have, for a number of reasons. Claims that single payer will save money amount to the assertion making government pricing more important to medical providers will make them less good at lobbying the government to get ever more money, which is a decidedly odd interpretation of how people respond to incentives.

            @herbert herberson

            the medicaid expansion only “worked best” for very specific definitions of worked. it “covered” more people. but most of those people were eligible for the pre-expansion program, they just hadn’t bothered signing up. Had they gotten sick, the first thing any hospital would have done is fill out the needed forms. it almost certainly didn’t make anyone healthier, at best, and evidence is mixed, it made a small number of people more financially secure, but we could have done that more efficiently by just giving them money.

          • Watchman says:

            Single-payer is a bad idea if you like accountability. In the UK our single-payer system (the National Health Service, always called the NHS to the point I question whether half my countrymen could tell you what the acronym stands for) is not that bad, but can fail spectacularly and as it has a tendency to prioritise staff over patients it can make solving problems or getting recompense very difficult. Admittedly this is because the payer (government) is also the provider, which is a worst case scenario to anyone outside a socialist mindset (hopefully), and the continental European systems tend to do better with more private provision. Where the UK is genuinely world-leading (a lot of political rhetoric in the UK likes to pretend the NHS is as a whole, but it is never backed by statistics) is generally in either areas where there is no private funding for research but where a state-run system can fund the concentration of expertise (so some cancer areas particularly) and in primary community care, which is significantly where doctors run their own practices and are contracted by the NHS to provide services, rather than being directly employed by the NHS.

            The best outcomes apparently come from systems combining single payer with insurance, which tallies with the UK system for dentistry (yes I know – insert stereotypical joke here and I’ll show off my not yellow or bent English teeth as I laugh at it) which is another top-performing area on the world metric,s and where unless you are in receipt of government benefits you have to pay yourself, but where prices can receive government subsidy.

            But even outright state-controlled single payer healthcare has to be preferable to the state-mandated, state-corrupted market that the US has. If government makes errors, at least it is accountable for them if it runs the health service. At the moment the federal government seems to be able to make systematic errors and then pretend they are the fault of someone else (not sure whether the population believes that mind you).

          • MrApophenia says:

            Who says central planning failed? The part of Obamacare that tries to leverage marketplace competition has fallen short of expectations, while the Medicaid portion has huge popularity and is responsible for most of the beneficial gains of the law.

          • Kevin C. says:

            I note that, vis-a-vis the debate about how single payer might work in the US, no one brings up the example of the VA.

          • cassander says:

            @MrApophenia says:

            Who says central planning failed? The part of Obamacare that tries to leverage marketplace competition has fallen short of expectations, while the Medicaid portion has huge popularity and is responsible for most of the beneficial gains of the law.

            A “market” where the government dictates what can be bought, what can be sold, and how much will be charged for it is not a market in any meaningful sense. And the assertion that the medicaid portion has “huge popularity” strikes me as ludicrous. I doubt that even a third of voters could articulate that the medicaid expansion is separate from the other ACA provisions, much less be strongly in favor.

          • Iain says:

            @cassander:

            One, actual single payer systems are relatively rare. Most countries use some sort of multi-payer system.

            Quibbling about exactly how to categorize individual systems is boring and I preemptively concede any such arguments. My point remains: compared to the American health care system, every other developed country spends less money, and every other developed country is more socialized and/or centrally planned. Do you think this is just a meaningless coincidence?

            Two, we don’t have to guess what a US single payer would look like, it has one called medicare. We know very well how much it costs, and that it has not been able to restrain costs the way other countries have, for a number of reasons. Claims that single payer will save money amount to the assertion making government pricing more important to medical providers will make them less good at lobbying the government to get ever more money, which is a decidedly odd interpretation of how people respond to incentives.

            Medicare and Medicaid already do spend less than private insurance. The idea that a monopsony can push down prices is literally Econ 101. If the harms of lobbying were greater than the benefits of increased bargaining power, then you’d expect Medicaid and Medicare to pay better than private plans. Equivalently, you’d expect the UK to pay higher rates than the US, because the NHS is such a juicy target for lobbyists. Neither of those is true.

          • cassander says:

            @Iain says:

            point remains: compared to the American health care system, every other developed country spends less money, and every other developed country is more socialized and/or centrally planned. Do you think this is just a meaningless coincidence?

            One, the argument that they are “more socialized” or centrally planned is questionable at best. Medicare engages in literal soviet style price setting by executive committee, and it’s a huge share of US medical spending. The rest is massively distorted by extremely large subsidies distortions, and regulation. The US is differently socialized, but it isn’t less socialized.

            Two, they have different political systems, with different dynamics, that produce different outcomes. The US does not lack for healthcare spending, it definitely lacks for evidence that the US government spending can reduce it.

            Medicare and Medicaid already do spend less than private insurance. The idea that a monopsony can push down prices is literally Econ 101.

            Econ 101, but not poli-sci 101. Using monopsony power to lower medicare spending has been tried and and it failed, utterly. You can’t say “we’ll use market power” when the medical lobby has demonstrated its ability, repeatedly, to convince the politicians not to. The same thing that happened with passing the ACA will happen with any single payer legislation, every provider will demand their pound of flesh. Pretending that this won’t happen for single payer isn’t just ignoring politics 101, it’s ignoring history.

          • Winter Shaker says:

            Siderea has a long but interesting series of posts about the proliferation of healthcare costs starting here. Difficult to boil down to a simple paraphrase, but the kernel of the argument is in part two:

            My hypothesis is this: that two things happened.

            The first thing is that the expenditures on health care began to escalate exponentially as a function of the increased health care available to buy, and this process, which had been slowly gathering steam through the 19th century and into the 20th started rounding the curve of the hockey stick in the 1960s and 1970s.

            Which brings us to the second thing that happened: the response. Just like in The Beer Game, players in the game reacted to the surge in demand, by attempting to do things to reduce costs. Wrong things. Precisely the wrong things.

          • Brad says:

            It’s awfully convenient that real concrete health care systems in other countries are completely irrelevant because of American anti-exceptionalism, but True Free Market ™ health care is obviously superior, will definitely work, and have no insurmountable implementation problems even though there are zero examples to point to.

          • cassander says:

            @Brad says:

            It’s awfully convenient that real concrete health care systems in other countries are completely irrelevant because of American anti-exceptionalism, but True Free Market ™ health care is obviously superior, will definitely work, and have no insurmountable implementation problems even though there are zero examples to point to.

            Not nearly as convenient as ignoring concrete examples of the desired system IN AMERICA repeatedly failing in exactly the ways predicted. But feel free to explain how I’m wrong. If the doc-fix happens with current medicare, why won’t it happen with medicare for all? What will magically change?

          • Edward Scizorhands says:

            I’d be surprisingly fine with single-payer (surprisingly to me, I mean, compared to my attitude 10 years ago). But it involves a number of trade-offs that its loudest proponents never talk about, and attempting to implement it without those trade-offs will lead to a much worse system.

          • But the Obamacare mess looks to me like an argument against (b), at least by demonstrating that an obvious alternative to single payer doesn’t seem to have worked out too well in practice.

            From the standpoint of an economist, the obvious problem with Obamacare is that it creates massive adverse selection. Insurance companies are required to overcharge young people and undercharge old people, to overcharge people initially healthy and undercharge people initially unhealthy. That makes it in the interest of healthy young people to be uninsured, and to prevent that it is necessary to compel people to buy insurance (or try to compel them) even when it isn’t worth what it costs to them.

            It was an attempt to combine health insurance with a massive income redistribution–from healthy to unhealthy and from young to old.

            There may be lots of other things wrong with it, but that’s the obvious problem from the standpoint of ordinary economic analysis, as per Akerlof’s classic “Market for Lemons” article. It’s as if people were required to buy and sell all used cars of the same age and model at the same price, independent of their condition.

          • baconbacon says:

            (both per capita and, I think, as a percentage of GDP)

            Percent of GDP is misleading. Government spending is a part of GDP, and so boosts it. US healthcare spending as a % of household consumption is much closer to other countries than % of GDP.

          • Witness says:

            @Winter Shaker

            Thanks for that Siderea link, it’s very interesting.

            I’m having the mind-boggling (and happy) experience this year that my insurance premium is going down this year – I have a Kaiser Permanente Plan (through my employer) with no deductible.

            I was skeptical last year when I switched over from a more expensive Aetna plan with a deductible that was (for me) becoming excessive. But the care has and coverage has been decent and (more importantly) we can afford to use it when we need it.

            Anyway, I bring it up because I suspect a lot of the reason that plan is cheaper is because they operate their own facilities and are therefore more mindful of the real costs of adding layers of paperwork/communication/micromanagement.

            I’m hopeful that this sort of thing can lead to competitive pressures that force the other “organ-isms” in the ecosystem to adapt (or perish), but of course I don’t have a ton of practical knowledge about how that might happen. (And also it doesn’t solve the other issue identified by Siderea – there’s still going to be ever-more treatments that we want to buy, so it’s likely we’ll keep spending more). Any thoughts?

          • Jaskologist says:

            Who says central planning failed? The part of Obamacare that tries to leverage marketplace competition has fallen short of expectations, while the Medicaid portion has huge popularity and is responsible for most of the beneficial gains of the law.

            The “success” here was the usual centrally planned systems are successful: goosing a specific metric without dealing with the actual problem. The number of enrollees went up, that is true.

            Health? That got worse. Life expectancy has gone down since Obamacare went into effect, and we’re dying sooner from a wide variety of causes.

            One of the many problems with bringing these systems into the political realm is that once it tribalizes, half the country loses its incentive to make the system work in the real world rather than on cherry-picked metrics.

          • more that the implementation had to keep every interest group in the medical industry happy.

            Thats seems to be a euphemism for “making a profit”. And no, you probably can’t create an efficient system under that constraint, but it’s not an argument against comprehensive public healthcare in general.

            Maybe a good start would be to call corruption corruption.

  16. ilikekittycat says:

    “Universe shouldn’t exist, CERN physicists conclude” is a ridiculous and confusing headline for “Scientists confirm CP symmetry violation”

  17. skybrian says:

    It’s weird how the Whitefish scandal blew up. Most articles are written by reporters who seem to have very little curiosity about basic questions: was Whitefish doing a good job or not at restoring power? And are they overcharging for it?

    It seems like reporting this story properly would require learning a lot more about how the power industry normally works and the logistics involved in restoring power in Puerto Rico. Instead we read about tenuous connections to Trump and editorials assuming it’s all very corrupt with little direct evidence.

    My pet theory based on following this story for a while is that PREPA is broke (they went bankrupt last summer) and they’re trying to conserve cash in unorthodox ways. They say they went with Whitefish because they didn’t require money up front and they didn’t need to provide lodging for the workers. Also, at one point they apparently were hoping the Army Corp of Engineers would do all the work for free. Going through normal channels, the mutual aid between power companies is not free and apparently they thought they couldn’t afford it.

    It seems like this might be more about people working at a company in a financially desperate situation making strange deals that a better financed company wouldn’t make. But nobody seems to be covering that angle.

    • Steve Sailer says:

      Puerto Rico is quite corrupt overall, but that’s not particularly interesting to the American press unless it can be somehow tied to Trump. For example, Puerto Rican public schools have unbelievably awful test scores on a Spanish-language version of the federally NAEP specifically made up to be culturally sensitive to Puerto Rico. But nobody in the 50 states much cares at all.

      After all, what’s the worst that can happen due to the ineptitude and corruption of government in Puerto Rico: the island continues to empty out, Puerto Ricans continue to move to Orlando, and Florida’s 29 Electoral Votes permanently switch to the Democrats? That sounds like a plan …

      • OptimalSolver says:

        Puerto Ricans continue to move to Orlando, and Florida’s 29 Electoral Votes permanently switch to the Democrats? That sounds like a plan …

        Everything sure looks like a nail, doesn’t it?

        • Steve Sailer says:

          Try Googling

          Puerto Rico test scores

          and see how much media interest there is in these spectacularly bad test scores.

          • albatross11 says:

            Steve:

            Is there good data anywhere about how the PR kids do when they move to mainland schools?

          • Steve Sailer says:

            Only 6% of Puerto Rican public school students score above the lowest of the four levels, Below Basic. Hispanic public school students in the US score about an order of magnitude better.

            23% of students attend private school in Puerto Rico, compared to about 10% in the 50 states, so that’s one reason, but …

            To focus on Puerto Ricans in the US, you can look at Hispanic NAEP scores in, say, Connecticut, which is heavily Puerto Rican.

            Puerto Rican parents in the Northeast are notorious for yanking their kids out of school when the weather turns cold and then showing up again in the spring. I don’t know if these snowbird students get tested on the NAEP or where.

        • Deiseach says:

          Everything sure looks like a nail, doesn’t it?

          Given that Bernstein’s 1957 “West Side Story” has lyrics in a song about the constant (and heavily implied to be unsustainable) population growth of Puerto Rico, its poverty and backwardness compared to the USA, and how the best thing that could happen is for the Puerto Ricans to emigrate to the mainland, I think we can say that this is a liberal attitude that has been around for a while: get with progress, leave that backwards island behind, come to the real America and become real Americans and leave those cultural habits that drag you down behind you.

          In fact, the original lyrics were so disparaging of Puerto Rico that the song was re-written for the 1961 movie version to include criticism of US attitudes to immigrant Puerto Ricans. And I don’t think anyone is going to maintain that Sondheim and Bernstein were solid Republican conservative Red Tribers, are they?

          ROSALIA
          Puerto Rico,
          You lovely island . . .
          Island of tropical breezes.
          Always the pineapples growing,
          Always the coffee blossoms blowing . . .

          ANITA
          Puerto Rico . . .
          You ugly island . . .
          Island of tropic diseases.
          Always the hurricanes blowing,
          Always the population growing . . .
          And the money owing,
          And the babies crying,
          And the bullets flying.
          I like the island Manhattan.
          Smoke on your pipe and put that in!

          OTHERS
          I like to be in America!
          O.K. by me in America!
          Ev’rything free in America
          For a small fee in America!

          ROSALIA
          I like the city of San Juan.

          ANITA
          I know a boat you can get on.

          ROSALIA
          Hundreds of flowers in full bloom.

          ANITA
          Hundreds of people in each room!

          ALL
          Automobile in America,
          Chromium steel in America,
          Wire-spoke wheel in America,
          Very big deal in America!

          ROSALIA
          I’ll drive a Buick through San Juan.

          ANITA
          If there’s a road you can drive on.

          ROSALIA
          I’ll give my cousins a free ride.

          ANITA
          How you get all of them inside?

          ALL
          Immigrant goes to America,
          Many hellos in America;
          Nobody knows in America
          Puerto Rico’s in America!

          ROSALIA
          I’ll bring a T.V. to San Juan.

          ANITA
          If there a current to turn on!

          ROSALIA
          I’ll give them new washing machine.

          ANITA
          What have they got there to keep clean?

          ALL
          I like the shores of America!
          Comfort is yours in America!
          Knobs on the doors in America,
          Wall-to-wall floors in America!

          ROSALIA
          When I will go back to San Juan.

          ANITA
          When you will shut up and get gone?

          ROSALIA
          Everyone there will give big cheer!

          ANITA
          Everyone there will have moved here!

          • cassander says:

            It’s almost certainly a good idea for the puerto ricans to emigrate, but is it good for the places the emigrate to?

          • The Nybbler says:

            Always the coffee blossoms blowing . . .

            You know the old joke about Communism in the Sahara resulting in a sand shortage? Well, Puerto Rico is so badly run that there’s been a Puerto Rican coffee shortage for many years. Which is too bad, because it was really good coffee.

        • vV_Vv says:

          More charitably, since the residents of Puerto Rico are in the unusual position of being US citizens who are not represented in the federal government, at least not until they move to the mainland, the federal government, and by extension any institution who has a stake in influencing it (read, the mainstream media) has little interest in their well-being.

          • Randy M says:

            Well, that’s more charitable to the Puerto Ricans, but on net probably more cynical, which is not to say wrong.
            One way or another, we really ought to fish or cut bait with that territory.

          • Evan Þ says:

            @Randy M, sure, as soon as they decide whether they want independence or statehood. In the meantime, I don’t have any ethical problems with continuing the current situation, since they seem to want it.

    • albatross11 says:

      Yeah, it seems like there’s a general problem I face as a media consumer these days:

      a. Trump really is an unusual president, with plenty of genuine missteps and disturbing comments and asinine behavior. It’s entirely plausible that he will wreck important institutions with his high-variance behavior.

      b. Nearly everyone in the prestige media *despises* Trump, and stories about Trump are instant clickbait, so the stories about him tend not to be overly careful to be accurate. There’s an automatic narrative everyone wants to write their story to, where Trump is uniquely horrible, even when he’s doing stuff every president before him has done or that absolutely any Republican elected to the presidency would have done.

      The result is that there’s a huge amount of chaff being spread around, making it harder to figure out when Trump or his administration are actually doing something seriously bad. To the extent Trump is inclined to do crazy or evil or destructive things, these stories provide him a lot of cover–people like me who have no love at all for Trump but also don’t think he’s Hitler 2.0 start tuning out the anti-Trump stories because most of them turn out to be nothingburgers.

      • Gobbobobble says:

        Yes, this is the exact problem I have. I really wish I knew of a media outlet I could trust to only bother with category a., but it seems they’re all either getting on board the tide of madness or shutting down in the face of it.

      • cassander says:

        a. Trump really is an unusual president, with plenty of genuine missteps and disturbing comments and asinine behavior. It’s entirely plausible that he will wreck important institutions with his high-variance behavior.

        One could, and I know many people who did, say the same thing about Obama, and make a decent case for it. See, for example, that this was reasonable satire. I think point B matters a hell of a lot more. Every president makes plenty of genuine missteps and can wreck important institutions with their behavior. Trump is not different in this regard.

      • Nornagest says:

        Same. I’m no Trump fan, I didn’t vote for him, I don’t like him, but it’s gotten to the point where the average Trump story (at least, of those that come up in my Facebook feed) is better evidence for the source being sensationalized to the point of being totally untrustworthy than for Trump doing anything unusually bad today.

        These days I mostly just block any sources that do it, because I’ll hear about anything really bad here on SSC.

        • pdbarnlsey says:

          I think it’s worth mentioning that the story is being framed not as “corrupt Trump government hires oddly small firm to fix PR’s power grid”, it’s “corrupt Trump government hires oddly small firm which happens to be exclusively based in the Secretary of the Interior’s tiny home town and to have employed his son to fix PR’s power grid”.

          That’s a big enough coincidence that it probably warrants the media asking questions, and it’s weird to me that not one of the posts in this thread even mentions it.

          • Nornagest says:

            I was responding to albatross11’s remarks about the general media climate more than to Skybrian’s about the Whitefish case. As you’ve phrased it, that does seem odd.

            But we’re still talking breadcrumbs, not a smoking gun. And if those breadcrumbs lead to actual corruption and not just run-of-the-mill incompetence (always a good bet!) or even genuine mitigating circumstances, that just makes your case one of the bad ones, which I did say or at least imply were out there. I don’t think that excuses the media climate; crying wolf is obnoxious at the best of times, but if there’s any chance of an actual wolf attack then it rises to reprehensible.

          • Watchman says:

            Isn’t the inherent logic of this narrative that government can only hire large firms that are already known to us, which might not exactly be a great way of getting the best deal. So hiring a small firm, which might still have the expertise and the ability to do the job (you don’t need people on payroll – you need to know people) wouldn’t be an issue.

            The link to the Secretary of the Interior is however the sort of thing that looks bad. But does the federal government not have procedures in place to get around this sort of issue? Because if the US cabinet can just award contracts, then the lack of previous scandals of this type is itself the major story.

          • Brad says:

            I’m not sure what a two person company can possibly bring to the table. If a two person company can do whatever it is, why can’t the government do it directly?

            With a 10,000+ person company they have already assembled significant capacity, internal mechanisms, and relationships that the government would have to recreate from scratch and so maybe it makes sense to just use what’s already there. Not so for a two person company.

          • Watchman says:

            I’m not actually questioning that there is something unusual here, but a two-person company can theoretically deploy as much equipment and personnel as a ten-thousand man company and has lower overheads, so is economically speaking much more efficient as a business model and maybe capable of offering a cheaper product – the internal markets of large engineering companies are not notoriously good ways of getting value for money for customers anyway.

            And there is one thing to check – are we sure this “two-man operation “don’t have the expertise. They are new (2015 according to their webpage), but have capital backers and apparently experience, so this is not just a small-town operation. Also they seem to be specialised in difficult situations – the ten-thousand man company might have a sub-division for that, but it is going to be ten men of the whole, so the scale difference drops away…. Presumably they will be using local labour for much of the work anyway (because its cheaper and politically more sensible).

            So another way to look at this (ignoring the personal link to the Secretary of the Interior) is that this is a specialised company’s first government contract, acquired two years after their foundation. It seems rather unreasonable to expect them to have a prior record of contracts in this situation. The image they present seems to fit the profile of a company that would be appropriate for this work, and the fact they have capital backers suggests they are looking to work at this scale, not doing grid connections for individual houses or whatever.

            The reaction shown here actually seems to show inherent bias towards existing companies and against new market entrants (a very bad idea considering how much of the economy and against location in small towns, but frankly for a company specialising in a fairly niche global market, why is basing yourself in rural Montana any worse than Washington – and it is a hell of a lot cheaper.

            Other things that make me think this is actually an honest organisation rather than a Republican’s cronies – they are hiring staff through the International Brotherhood of Electrical Workers, so clearly work with unions (not surprising if the founder is an ex-union member); they specifically say they have extensive subcontractor relationships (I would suggest their major asset that attracted the capital investors before the government contract was probably the address book); they look like a start-up by an industry veteran, not a local power supply outfit.

            There still could be an issue of favours for a local company here, but the company themselves do not appear to be the wrong people for the job. Upthread somewhere someone comments on how reporting of Trump adminstration actions tends to portray everything through a very distorted mirror – I think we might have a case in action here.

          • John Schilling says:

            but a two-person company can theoretically deploy as much equipment and personnel as a ten-thousand man company and has lower overheads,

            They can’t deploy teams, and the difference between a team and a collection of skilled people who have never worked together before is huge. Far bigger than the overhead costs of a company that has ten thousand people already organized into project teams.

            And they can’t deploy the manpower of a ten thousand man company fast, because do you have any idea how long it takes to hire ten thousand people if it’s just you and your partner? Granted, the first thing you do is hire people to help you with hiring more people, but that still doesn’t happen in days or weeks.

            Both of these work better if you subcontract to existing corporations with teams of people already on staff, but now you need negotiators and lawyers to set up all the contracts, and that doesn’t happen fast either. Particularly where government contracts are concerned, and government contracting rules don’t stop with the prime contractor.

            If you limit it to subcontracting with a few big companies, sure, that can be set up reasonably fast, but why aren’t we just going straight to the big companies?

            Rebuilding Puerto Rico’s electric power infrastructure is something that really needs to be done ASAP. This is not the right project for a new company to show what it can do, not a project where someone who doesn’t have the resources (human or otherwise) can compete with someone who does by saying “I’ll buy the resources”. You are working from a model of How To Do Business Better Than The Government Usually Does Business, that may have something to say for it in the general case but is not appropriate for this specific case.

            And whoever gave the contract to these two guys, I am fairly confident has enough experience in government contracting to understand that. No matter how much experience these two guys have in their previous career, this stinks of corruption.

          • baconbacon says:

            I’m not sure what a two person company can possibly bring to the table. If a two person company can do whatever it is, why can’t the government do it directly?

            Because what is needed is change, on a fundamental level. Economic growth occurs because of new approaches (Kling’s PSST) not because your water wheels are 1-2% more efficient than they were last year. The latter approach automatically caps out, and probably faces diminishing returns before then. Instead of a water powered mill you have an electric mill hooked up to a hydroelectric dam. “That will never work” cry the Millers, “you will never profit from all that expense!”, but eventually it is realized that the a major expense, the a major limiting factor, in farming was being constrained by access to mills that had to be built at specific locations.

            What can a 2 person company do? Innovate, start from scratch and build a daring new paradigm without risking a multi-billion dollar company in the process without having to deal with corporate (or worse government) inertia.

          • Iain says:

            What can a 2 person company do? Innovate, start from scratch and build a daring new paradigm without risking a multi-billion dollar company in the process without having to deal with corporate (or worse government) inertia.

            Cool. Maybe they could start testing their “daring new paradigm” in a less time-sensitive situation where fewer lives are at stake?

          • baconbacon says:

            Oh, its time sensitive and lives are at stake? Well then Justice Friends, assemble and quickly fill out your RFPs, make sure to document everything because once we get through the first round of bureaucratic nonsense there are still 3 more levels, but then it will be Justice Away!

          • Brad says:

            What can a 2 person company do? Innovate, start from scratch and build a daring new paradigm without risking a multi-billion dollar company in the process without having to deal with corporate (or worse government) inertia.

            Without risking a multi-billion dollar company, but in this case risking $300 million in government money and ending up with the problem still in place and needing to be fixed by more traditional means. If some government entity has that kind of risk tolerance then why not try the daring new paradigm themselves?

            I can’t see any reason why we shouldn’t have a strong presumption of corruption in a case where a two person company gets a $300 million contract. They would have to personally have some amazing skills that couldn’t be purchased any other way for it to make any sense. Because two people’s skills are the only thing they bring to the table. Everything else has to be bought or hired with money coming in from the contract.

            It is a little like an aquihire but you aren’t even getting the team under your direct control.

          • John Schilling says:

            Oh, its time sensitive and lives are at stake? Well then Justice Friends, assemble and quickly fill out your RFPs, make sure to document everything because once we get through the first round of bureaucratic nonsense there are still 3 more levels, but then it will be Justice Away!

            If you need things done quickly and are willing to do away with things like competitive bidding and audits, then the way you do that is to find an existing organization with the resources to do the job and give them the sole-source no-bid contract. I don’t know who the equivalent of C.C. Meyers is for power grids, but I’m about 99% certain it isn’t two guys in Montana.

            Nobody is arguing that the government should never hire two clever, hard-working guys with the right idea and the freedom to innovate. I’ve been one of those guys. But arguing that this is the right paradigm here, even that it is tolerably close to right, I’m not buying it and you’re barely even trying to sell it.

          • baconbacon says:

            Nobody is arguing that the government should never hire two clever, hard-working guys with the right idea and the freedom to innovate. I’ve been one of those guys. But arguing that this is the right paradigm here, even that it is tolerably close to right, I’m not buying it and you’re barely even trying to sell it.

            I’m not selling it because there is no point in selling it. The issues in PR or NO after Katrina aren’t about the mobility of the Federal Government, or the decision tree or the size of the organization running the relief efforts, they are about entrenched poverty with the trappings of wealth. The responses of FEMA or whoever are generally in line with restoring and maintaining the status quo, but for these areas the status is not quo. When you are at a point where the argument is entirely based on how to best return to a broken system you have missed the point, missed the ability/opportunity to create a robust/dynamic system.

    • Deiseach says:

      Things like the linked Google and Tesla, and Pitbull sending his private plane (I know this becuase it’s all over my feckin’ Tumblr dashboard and of course compared to Trump’s disfavour) and the rest of it are publicity stunts, even if well-meant, and only work because it’s for one defined task that is small compared with the task of fixing the entire rest of the island, and has wealthy backers who can throw money at it. I doubt Pitbull is paying for a fleet of planes running for this entire time, because he simply can’t do that.

      The mess about Puerto Rico probably does have a lot of “not enough help, corruption, inertia and indifference, too many government cooks, etc etc etc” involved, but saying “Why isn’t Government Response as fast, efficient, and magical as One-Off Small Project has no idea of the scale involved in getting public infrastructure back up and running, not to mention all the other problems.

      • albatross11 says:

        The lesson from both Maria and Katrina is that if you want to do well in a disaster in the US, you really need a competent local and state government. If you’re counting on the federal government to save you, you’re probably going to be sitting in the dark, hungry, for a long time.

        Louisiana and New Orleans have famously inept, corrupt governments. So does Puerto Rico. (I don’t know about San Juan.) The result is that when a big disaster struck, there wasn’t an effective response–they didn’t have the money set aside, or the planning done, or the infrastructure maintained, or whatever needed to happen.

        Exacerbating this: financial problems lead to deferred maintenance, and they drain your rainy day fund and make credit hard to come by. They often lead to getting rid of excess staff. All that tends to bite you in the ass when there’s a crisis.

        • Edward Scizorhands says:

          If you’re broke, you will be rescued, so at a certain point in poverty it becomes self-defeating to save for the future because your only plan is “look pitiful.”

      • poignardazur says:

        Which reminds me of the article where Scott was comparing Trump to Batman. The dynamic here is the same: they’re pouring a lot of resources into a very visible place to look generous and effective.

        Honestly, I starting to think that getting excited about companies doing good things is almost counter-productive. When companies do things efficiently, we don’t need to get excited about it, because they do it for money (I’m tired, not sure how much sense I’m making).

    • MrApophenia says:

      And are they overcharging for it?

      In fairness, the contract explicitly banned the government from asking that question, which seems pretty weird.

  18. suntzuanime says:

    From the link about how we all actually love tech companies:

    Companies are designing their products to get you to take actions that help their businesses.

    This is part of modern capitalism and it’s been around forever. Tech companies making notification buttons red instead of blue because red evokes a more urgent response are not fundamentally different from an advertisement that features happy beautiful people using a product, or the way grocery stores display products to guide you to the more profitable ones.

    Criticize modern marketing or free markets, but this is not anything specific to tech. (Except that they hire really smart people who are quite good at what they do.)

    It seems to me that really smart people who are quite good at what they do is something to be specifically concerned about, when what they do is steal men’s souls and enslave them. You can look at this mentality taken to its extreme in the realm of gatcha mobile phone games. Maybe there’s something more generally beneficial for us as a society to use our limited supply of really smart people who are quite good at what they do on than a war of attrition for consumer attention?

    • poignardazur says:

      I think / hope that the whole “steal men’s soul and enslave them through better target advertising” thing is anti-inductive. Advertising is getting better and more insidious as the same time as people are getting better at tuning it out / installing add blockers / not actively imagining themselves in place of the handsome happy person driving the expensive car. (except children, because they have to learn that shit from the ground up)

      When you think about it, early 20-Century customers already had to worry about predatory supermarket design, but they didn’t have price comparators.

      • Aapje says:

        Except for the people who are less capable to defend themselves. I’m not very happy with an outcome where an intelligent elite is successful at defending themselves from predatory marketing, while the rest of society is coerced into self-harm because they cannot see how they are being manipulated.

      • actinide meta says:

        Yeah, I remember flipping through an old magazine and finding the ads hilariously unsubtle. “Nothing makes a man more popular with the ladies like a steady job in accounting with a degree from BOB’S DIPLOMA MILL!” (Not a real quote, but in the style I remember.) I have to think that ads got subtler not just because advertisers got smarter but because their victims target audience did.

        That said, I think @Aapje’s concern about the most vulnerable being disproportionately hurt by this arms race seems pretty realistic. Now instead of paying money for a video game, you can play a pretty good game for free while some poor sucker who can’t afford it dumps his life savings into it to subsidize you.

        My hope is that someone can figure out a way to build a brand out of (the idea, not the slogan) “you are our customer, not our product!” And that once people came to believe in that they would pay for it.

        Also, do you ever get the feeling that people have in some other ways gotten more gullible/trusting? For example, I feel like there was a time when very few people would have been willing to accept impositions on privacy that are routine today, assuming that the other party would use their knowledge to take advantage of them. Maybe this is just the dark side of a “high trust” society.

        • skybrian says:

          I suspect the dynamic is that a few rich but foolish people (“whales”) dump a lot of money into it, and the game is built around attracting them. Compare with attracting “high rollers” in Vegas casinos and 419 scams.

          Or more generally, any kind of price discrimination (like the airlines do).

          There is something appealingly democratic about charging everyone the same, well-known price. However, arguably that’s less fair than attempts to get rich people to pay more. Maybe they (we) should pay more?

          • VivaLaPanda says:

            This is the case. I remember reading a study but it was a long time ago and I don’t have a link, so take it with a grain of salt.

          • actinide meta says:

            It would be nice to think that the people who wind up addicted to these games are all rich, but what are the chances?

            It’s true that a very small fraction of players are paying for these games, and therefore spending lots of money. But that doesn’t mean they can all afford it.

          • lvlln says:

            I too read an article a bit ago that said that mobile free-to-play games are basically funded by like 1% of players spending exorbitant amounts of money (with the top 0.01% spending exorbitant amounts relative to the other 0.99% within the top 1%). The F2P game company that they interviewed in the article even mentioned targeting specific whales, friending them on Facebook, learning their favorite sports teams, and creating new expensive in-game items whose color scheme matched that of that whale’s favorite team. I found that that level of manipulation was going on to be quite fascinating.

            On the one hand, such stark disparities where some players spend tens of thousands of dollars while (most) others spend literally nothing seem shocking. On the other, it strikes me that this should be the most obvious thing in the world when I think about it for more than 10 seconds. The Pareto Principle shows up again and again pretty much everywhere, particularly in issues of economics, so my default presumption of how the economics of F2P games work should have been something like that.

            I don’t have a link either, unfortunately.

            An aside, I’ve just recently gotten into a F2P game called Fate Grand Order. Its random draws are cruel, where when they have special high probability events, your probability of receiving the special top card everyone wants goes up to 0.7%. It costs about $3 for each draw, which means an average of over $420 per top card. To me, such a ridiculously high price to come anywhere close to a good chance at drawing a top card pushes me away from spending any $ at all, and I intuitively would have thought that they’d make more $ by raising rates/lowering prices, but clearly what they do works for them. And I’ve seen reports of at least one player who has gotten at least 5 of literally every top card, of which there are dozens. You don’t even need 50 players of that style to fund that game for the 2 years it’s been out.

        • suntzuanime says:

          That said, I think @Aapje’s concern about the most vulnerable being disproportionately hurt by this arms race seems pretty realistic. Now instead of paying money for a video game, you can play a pretty good game for free while some poor sucker who can’t afford it dumps his life savings into it to subsidize you.

          The problem I have with that is that the gatcha mobage aren’t even good! This isn’t a case where we’re walking away from Omelas because we’re worried about the poor sucker who’s paying for it, the games are designed to torment you into paying up, and then torment you more. At least in the case of the whales somebody’s getting paid, the poor souls with enough control not to pay money but not enough control not to play are getting tortured by an uncaring machine for nobody’s benefit.

  19. arancaytar says:

    Hint: don’t do what I did and use negative quantum operations to turn back time, it just makes you start over from the beginning.

    But it was the only way I could go back and accept the Drifters’ offer! The game would have been over and I wouldn’t have spent an additional four days playing it…

    Oh.

  20. nestorr says:

    Cruise ships: May be true, I obviously am unqualified to say. But it’s really hard to believe, I was just looking at one in port yesterday. They’re big, white and clean looking and there’s no visible exhaust at all when they move. Obviously we’re talking invisible particulates but psychologically they don’t look like heavy polluters.

    The volume figures are also a hard sell, I can’t fathom the 1 ship= 1.000.000 cars equivalence. You could fit 1.000.000 car engines in the volume of a cruise ship maybe but certainly not the volume of it’s engines… and I’d expect the ship’s engines to pollute less just in terms of surface to volume ratio, though that may be a naive way of looking at it.

    I assume the ship’s engines burn low grade fuel and in the microparticulate pollution stakes they’re worse than car engines by a factor of 1000 and then the volume covers the other 1000?

    • AlphaGamma says:

      In port: Which port? Some of them have started installing shore power connections for cruise ships specifically so they don’t have to run their engines/generators in port to generate power for hotel loads.

      As for numbers, you’re right about low-grade fuel. The ship in the article (MV Oceana) is powered by four Sulzer 16ZAV40S diesels, each of which has a total cylinder volume of 1120 litres. So across the four engines we have the displacement of a few thousand cars (depending on car size). Their power output is 46 megawatts or about 60,000 horsepower (again, 1000 small cars). The rest comes from differences in the fuel used.

  21. sustrik says:

    Here’s my personal story to support Siderea’s point about New Atheism: I was raised as an atheist and I have never had any business with religion. So, one could argue, New Atheism was preaching to a person who already got the message in this case. Yet, they’ve managed to influence me. Before, I haven’t though about being religious/non-religious as an ethical issue. If some of my friends were religious then it was weird and curious, but shrug, in didn’t matter in the end. Nowadays I feel about them similarly to how I feel about young communists in Che Guevara T-shirts: These guys may be a good people personally, but they should really spend some time questioning what they believe.

    • Aapje says:

      Nowadays I feel about them similarly to how I feel about young communists in Che Guevara T-shirts: These guys may be a good people personally, but they should really spend some time questioning what they believe.

      Sure, but that’s also how I feel about most people.

      • Great point Aapje. Most of humanity spends very little of their time and effort seriously thinking about their beliefs. Even most of those who purport to be intellectuals are just following tribal beliefs with little thought of their own. Theists are not any worse than any other group. In fact, atheists on average default to tribal beliefs as much as anyone else, in my experience.

  22. B_Rat says:

    “Universe Shouldn’t Exist, CERN Physicists Conclude” is a load of bullmanure, as sadly increasingly often in fundamental physics’s university press releases. The blog I linked to has a very populated category for that.

    • eyeballfrog says:

      To go into slightly more detail, the question of baryon asymmetry, i.e. why there’s more matter than antimatter, comes from violating CP symmetry. The existence of CP violations is well-known, but it’s not clear how the currently known ones led to the level of baryon asymmetry observed. They’re kind of weird and are bound up in obscure things involving the weak interaction, so finding new examples of CP violation is always of interest.

      However, the antiproton’s magnetic moment being different in magnitude from the proton’s would violate CPT symmetry. CPT symmetry is a consequence of special relativity, so a violation of that would mean special relativity is wrong. This would be a massive upheaval of most of modern physics, and obviously was not expected to happen.

      So this was less “scientists failed to find what they were looking for” and more “scientists found exactly what they expected”.

      • B_Rat says:

        Exactly, what’s worrisome is that while one usually castigates journalistic sensationalism in science news, at least in Physics in these years I’ve seen plenty of university press releases that were overhyped up to the preposterously misleading.
        Woit is of the opinion that String Theory’s failure, with the multiverse mania and such, has plenty to do with that, but I fear that it might have sociological roots too.

        Edit: Rigourosly, CPT theorem depends on local relativistic field theories’ formalism, but that’s not anything that’s really gonna be discarded anytime soon, too.

        • Deiseach says:

          Is it to do with funding chasing and justifying big expensive experiments? So you can show your backers a big splashy news story about WORLD-CHANGING RESULTS and that makes them feel happier about handing over a few billion or more to build a particle accelerator?

          I think half of the hype about stem cell research was done on this level, too; issuing press releases about CURING ALL DISEASE WITHIN TEN YEARS IF ONLY WE GET PERMISSION TO DO THIS (AND CHINA WILL DO IT FIRST ANYWAY) when privately they knew it was nowhere near that level, and would never be, and it was research that would help unlock some of the mechanisms of some conditions, not any kind of “this will cure Wilkins’ Wobbles Syndrome” at all.

          • Watchman says:

            No – it’s simply press stories are a metric used in universities to assess their impact (reputation is a major thing in universities, and is a factor in global league tables). So there are incentives to have your research written up and then misrepresented by science journalists who at best seem to copy and paste the press release, rather than reading the actual paper.

    • B_Rat says:

      I don’t know what went wrong, here’s the first link.

  23. t mes says:

    Love Huemer’s LOR article. Is there any chance at all that we see reforms in this area? What do non-traditional students who have been out of academia for 5-10 years do anyway? Overall seems like a very regressive, anti-outsider policy which is interesting since admissions boards are always touting how much they strive for diversity.

  24. jddt says:

    “Speaking of ACLU cases: city of Dickinson, Texas, tries to make hurricane relief aid conditional on recipients promising not to boycott Israel”

    Last time I checked this wasn’t true but was either mis-understood or maliciously made up. http://www.honestreporting.com/are-israel-supporters-to-blame-for-hurricane-harvey-victims/

    Since HaAretz broke the story (which subsequently turned out years later to be nonsense) about the Arab being prosecuted for having sex with a Jewish lady without making it clear he was Arab; and then the story that Israeli doctors were sterilising Ethopian Jews without their knowledge to control their population (which again turned out to be nonsense); my HaAretz rule of thumb is that if it sounds shockingly awful, it’s probably nonsense.

    • Evan Þ says:

      the story (which subsequently turned out years later to be nonsense) about the Arab being prosecuted for having sex with a Jewish lady without making it clear he was Arab

      What happened with that? I never heard the followup.

    • herbert herberson says:

      That link is only a partial rebuttal. The law exists, the forms exist, and nothing I read suggested that prosecutions had occurred (and when it comes to regulations touching on speech and ideology, selective prosecution opens as many cans of worms as full enforcement). All it really does is correctly point out that it only applies to contractors and pretend that the applicability of boycotts to the First Amendment is a fully settled issue.

      Interestingly, this story was all over the left-o-sphere a week or two ago and nothing I read then gave me the impression it was for aid recipients, rather than contractors, in the way the Haaretz piece incorrectly suggests.

  25. Soeren E says:

    Thanks for posting the link to the Map of the AI Safety Community.

    Counter-nitpick: I think Roman’s tower looks reasonably Roman. First result on Google Images for “Roman Tower” is this:
    http://www.gothereguide.com/Images/UK/Oxford/Roman_tower_oxford.jpg
    (From Oxford, the primary AI Safety city within the borders of the Roman Empire.)

    • Soeren E says:

      Roman Yampolskiy (from Riga, Latvia) pointed me to Sigulda Castle Tower, which was built by the (Holy) Roman Empire, is quite close to Riga, and seems to be of roughly the same style.

    • quaelegit says:

      Scott’s initial comment really confused me because Romanesque Architecture IS medieval buildings. (I can’t load the map right now so maybe it’s a particularly non-Romanesque-looking medieval building.) I don’t know what the proper adjective is for “in the style of Roman buildings of Antiquity”…

      • Watchman says:

        Depends on which period – there was a long time with Romans around, and they weren’t conservative about their architectural styles.

        • quaelegit says:

          Good point. For all I know, maybe the building on the map really does look like something the Romans built. (Since I can’t see the map and I don’t know Roman architecture outside of the Pantheon, aqueducts, Pompeii, and what’s currently preserved in the Forum…)

  26. I can’t believe Scott missed the chance to give this post a Tigger warning.

  27. beleester says:

    The RNC is crushing the DNC in fundraising, but Democrats appear to be giving directly to the candidates instead of to the DNC: http://talkingpointsmemo.com/dc/congressional-democrats-crushing-gop-in-fundraising

    • albatross11 says:

      I rather suspect that the DNC’s performance in the last election hasn’t filled a lot of Democrats or liberals with confidence that it will spend their money effectively in the future.

    • Deiseach says:

      Democrats appear to be giving directly to the candidates instead of to the DNC

      Which would make sense if they believed all the stuff spouted during the DNC Chair Mini-Scandal(s) about “This is no big deal! Nobody cares about the DNC! It has no power! Debbie Wassermann just wandered in off the street and we mistook her for the new cleaning woman so that’s how she got a job here! And it is of no importance at all who gets elected as new chair, so stories about this tearing the party apart are nonsense, stop writing them, the DNC is only a… thing… actually we have no idea what it does or what it’s for, so stop talking about it, IT DOESN’T MATTER!”

      I mean, if I’m a Democrat supporter and everyone in the party is telling me the DNC means nothing and does nothing, naturally I’m going to give my money directly to my local candidate and not some national party boondoggle.

      • BBA says:

        Well, it was mostly true. Sadly, not the part about DWS wandering in off the street, since a random person off the street could’ve done a better job than she did, but the rest of it…yeah.

      • In the U.S., national party committees have pomp and circumstance but almost no power. And over time, party organizations in general have been losing power, not gaining it.

        Political polarization might make it look like parties are gaining strength, but they’re not. Individual candidates are really on their own. It’s very different from parliamentary democracy.

        • An interesting general point. Do you have any conclusions about the consequences? Are there different failure modes for the different democratic models–strong parties vs weak parties?

        • Nornagest says:

          I’ve heard it phrased as “partisanship is strong, but parties are weak”.

          (Don’t remember where, though. Google turns up several seemingly-independent usages.)

          • I’ve heard it phrased as “partisanship is strong, but parties are weak”.

            I wish I had come up with that!

            Do you have any conclusions about the consequences? Are there different failure modes for the different democratic models–strong parties vs weak parties?

            Conclusions? No.

            Certainly it’s a remarkable contrast in political culture between the U.S. and any random European or Asian democracy. I don’t think it could easily be changed.

        • Deiseach says:

          There’s a marvellously click-baity headline to this story, and the opening is equally as overwrought, but what do you think of this, in the light of the claim that the national party headquarters means damn-all?

          https://www.politico.com/magazine/story/2017/11/02/clinton-brazile-hacks-2016-215774

          I love the start here; sweet mother of mercy, girl, you’re only going to make a phone call not perform open-heart surgery! And what’s with the quasi-religious ritual anyway?

          Before I called Bernie Sanders, I lit a candle in my living room and put on some gospel music. I wanted to center myself for what I knew would be an emotional phone call.

          It’s interesting to read a Bernie stan point of view, as to date I’ve only seen the Hillary devotees – this woman dares criticise the Sainted Barack, gasp!

          My predecessor, Florida Rep. Debbie Wasserman Schultz, had not been the most active chair in fundraising at a time when President Barack Obama’s neglect had left the party in significant debt. As Hillary’s campaign gained momentum, she resolved the party’s debt and put it on a starvation diet. It had become dependent on her campaign for survival, for which she expected to wield control of its operations.

          I do love the American New Journalism prose style, though, with all the “I gasped”, “I screamed”, “it broke my heart” throughout this – it makes it sound like a Gothic romance novel! 🙂 This is golden – here, our love-lorn heroine makes an agonised decision:

          I had to keep my promise to Bernie. I was in agony as I dialed him. Keeping this secret was against everything that I stood for, all that I valued as a woman and as a public servant.

          The woman’s heart says “No, no!” but the stern duty of the public servant demands “Yes!”

          It’s a fun read, even if I don’t know if I believe any of it. Well, apart from the Clintons turning the DNC into Hillary’s campaign piggy-bank, that seems to be borne out from other articles I’ve read previously.

          • Standing in the Shadows says:

            Everything I’ve read about HRC’s impact on her support organizations, and even her impact on own self and her own plans, appear to be great examples of the old adage:

            “Every organization can be accurately modelled as if it had been taken over by a secret cabal of it’s own worst enemies.”

            She is the worst thing that ever happened to her husband’s legacy, to her party, to her employers, and to her circles of followers, supporters, enablers, aids, employees, and staff.

          • @ Deiseach

            but what do you think of this, in the light of the claim that the national party headquarters means damn-all?

            They’re all playing in the sandbox. OMG, Debbie took out loans without telling the party officers! Holy crap, the Hillary campaign was using the DNC as one of many pots to slop money around! Wow, Bernie was never told what was going on!

            The party national committee is a compliant vehicle for the incumbent president or the party’s nominee. The Obama White House didn’t need the national party, so it became irrelevant, and withered. It was all too grateful to accept a lifeline from someone with an agenda, who they assumed would soon be the nominee anyway.

            Does that sound like a dynamic, powerful organization that has influence in the real world?

  28. Rachael says:

    The MBTI link doesn’t seem to work (imgur having technical difficulties). I’ve tried a few times over a couple of hours.

  29. FeepingCreature says:

    I never understood why the universe had to be asymmetric. Can’t it just be symmetric without annihilating itself? Ie. can, for instance, all the remaining antimatter be off beyond our Hubble horizon, forming an identical subset-universe with the opposite sign?

    • DocKaon says:

      That would certainly be a solution, but then you’d have to explain how the separation of matter and anti-matter occurred. How and why did such a separation occur in the early universe? Which of course would imply some sort of phenomena which treated matter and anti-matter differently, i.e. a matter anti-matter asymmetry.

      • Deiseach says:

        you’d have to explain how the separation of matter and anti-matter occurred. How and why did such a separation occur in the early universe?

        Irreconcilable differences.

      • FeepingCreature says:

        Could it be a quantum event? Ie. there was some process going on across the boundary between the two symmetric universe halves that had two outcomes with opposite signs, and both happened, and we happen to find ourselves in the universe where the sign came out “matter”, but it’s still absolutely symmetric, just indexically asymmetric?

    • B_Rat says:

      The fact is that one should either propose a mechanism for that or suppose that it happened by chance. In the latter case, someone should compute a meaningful probability of it happening and of us not seeing the border between sole-matter and sole-antimatter “universes”, in the former the second problem holds: in any case, nobody is really sure even about how such probabilities should be defined, so the idea on the whole isn’t too much falsifiable and it’s regarded as pure conjecture.

      Having an explanation like “Of course, even if you start with the same number of particles and antiparticles symmetry violations cause you to end up with more particles” would be much more neat (whether known symmetry violations suffice is debated, though most are of the opinion that they do not).

  30. keranih says:

    So, regarding the work camps vs prison…there is so much that is apparently being done wrong at those institutions, but the parts of that article that I *do* know about are slanted, stated in an inflammatory manner or just plain *wrong* that I’m hesitant to credit the article for anything accurate on the rest.

    Also, it would be a little funny to me, who is very frustrated with the ACLU for (currently) flirting with taking yet another amendment off its “liberties to be cherished” list, to have the ACLU force a court case that comes down and says “yes, these work camps are bad. Everyone who might have gone to one for one year must instead go to prison.”

    Because nothing painful, traumatic, or life altering happens to people in prison.

    • Winter Shaker says:

      To be fair, the ACLU is also against the policies that cause people whose only crime is to use a drug to be sent to [work camps / prison] in the first place. After all, if you get caught using a drug, and are faced with a choice between rehab and jail, you’re going to prefer rehab even if there is no way that your use of the drug would have been problematic enough to justify rehab in the absence of a system that sees it as inherently evil to enjoy certain altered states of consciousness.

      It is possible that the ACLU will succeed in chipping away a small part of the whole travesty-of-justice that is is the War on Drugs while failing to eliminate it entirely, in a way that leaves things worse, but I don’t expect them to stop trying to argue the case.

      • Jiro says:

        Someone pointed out relatively recently, either here or in the Reddit thread, that most criminals are in jail for violent crimes, not drug crimes, and that a common statistic about drug criminals in Federal prison is misleading because most criminals are in state prisons.

        • Winter Shaker says:

          Sure. But this particular case was specifically about drug users being sent to either ‘rehab’ or prison, rather than the broader class of criminals-in-general. I would put it past someone to try, but I’d expect it to be harder to spin it as a worthwhile project to divert, say, muggers or burglars (with no known drug connection) to a work camp that is specifically billed as a drug rehab programme.

          Edit: I originally commented on the basis of having only read Scott’s precis. It appears I was insufficiently cynical, having now clicked through to the article, in which the first person profiled was indeed sent to ‘drug rehab’ for an entirely non-drug related crime. Oh well…

          • albatross11 says:

            It’s a really bad policy, in general, to have to police or prisons operating both for public safety and also as a source of revenue. I don’t know the specifics of these drug rehab programs, but I think it’s a terrible idea to have your prisons (or mandatory drug treatment centers) making a lot of money for someone. That’s a perfect setup for corruption.

          • Kevin C. says:

            @albatross11

            It’s a really bad policy, in general, to have to police or prisons operating both for public safety and also as a source of revenue.

            Yes, but on the other hand, I’ve seen plenty of left-leaning folks praise the use in other countries of penal labor as an alternative to (American-style) long jail sentences.

          • pdbarnlsey says:

            I’m not sure that’s true, Kevin, though “plenty of left-leaning folks” is a tough category to adjudicate.

            If you mean Singapore-style public garbage collection in lieu of imprisonment(/corporal punishment), that’s typically classified by the criminal justice community as “stigmatic shaming”, and the state-of-the-art punishments (circa 1999 anyway) lean more towards “re-integrative shaming”, though this might end up being a post-hoc distinction in some cases.

            There’s a strand of cultural relativist leftism which tends to praise, or at least defend from criticism, anything which other cultures do. They’d probably be in favour.

          • Kevin C. says:

            @pdbarnlsey

            If you mean Singapore-style public garbage collection in lieu of imprisonment(/corporal punishment)

            Actually, I was thinking more like Dutch taakstraf.

          • pdbarnlsey says:

            That’s just a community service order, Kevin. These are extremely common in developed nations as an alternative to jail for minor crimes – I would have thought universal. Does the US not have them?

            But, yeah, if that’s what you meant by “penal labour” then I’d say you are correct that they are popular with most of the left – rightly, I’d say. It’s worth noting that they are not combined with imprisonment, though, they’re an alternative to it.

          • Nornagest says:

            Does the US not have [community service orders]?

            Community service is a pretty common sentence in the US, although I don’t know how it compares to the Dutch version. Usually imposed on first-time offenders for minor crimes, or as a condition of parole.

    • herbert herberson says:

      Because nothing painful, traumatic, or life altering happens to people in prison.

      The country is chock-full of drug courts who don’t use meat processing plants for rehab. Shut down this particular practice, and it’s at least as likely that it will be replaced by a real drug court as it will by nothing–plus, it reduces the chances that other entrepreneurs will try to replicate the practice.

      Curious what you found slanted. I have some familiarity with this world and didn’t see any red flags.

  31. Jeremiah says:

    Scott gave me permission to do a podcast version of SSC, which is now up to nine episodes. I keep hoping he’ll put a link to it in a post like this, but, understandably, he doesn’t want to dilute his brand with my nasally readings of his posts. (In reference to a previous comment on another thread Tom Waits was unavailable, and if it were completely up to me I’d have George Guidall do it). In any case here’s the iTunes link:

    SlateStarCodex Podcast

    Finally, I assume that there’s no point in making an audio version of a link post, since because the whole point is being able to follow the links, but if someone thinks otherwise, let me know. It wouldn’t be too much trouble.

  32. fortaleza84 says:

    I live in the NYC area and if I see a middle-aged Italian-looking guy driving an SUV, it occurs to me that the guy’s name is “Tony” are a good deal higher than chance expectation. But what’s interesting to me is that you can take out age, ethnicity, and class markers and still get a significant effect. I’m actually pretty skeptical of this and would guess that the study designers are not properly accounting for age, ethnicity, and class.

  33. themadmammaker says:

    Those MBTI facial phenotypes look mighty suspicious to me – some of them are smiling, some are not, I find it hard to believe that people of different personalities would have *such* differences in facial expression.

    That looks like someone made a mash of like 4 or 5 faces with similar expressions and then assigned the labels to them post-hoc or something.

    • eyeballfrog says:

      According to the link they took 36 faces of each of the 16 types for each gender and averaged them.

      What’d be nice is if they presented it in a grid.

  34. VirgilKurkjian says:

    I have the good fortune to be at an institute with some prominent AI critics. I’ve spoken to a few about AlphaGo and their opinion (echoed in some comments above) is that while the algorithm may appear to learn a great deal from very little, huge amounts of information have actually been encoded in the expectations of its architecture, and the team hasn’t done enough to show that 1) the results of this system on learning Go are robust to variation in the parameters, or 2) the parameters need to be precise but they are not “tuned” to Go, and can learn/perform equally well in a number of other domains. Short of this sort of evidence it’s not much of an advancement and not terribly exciting.

  35. Peffern says:

    In light of Tesla helping to rebuild Puerto Rico’s grid, Amazon headquarter shenanigans, and various things Facebook is doing, I can’t be the only one whose mind keeps pattern-matching to Snow Crash.

    Am I just paranoid?

    • Bugmaster says:

      Nah, I see it as something closer to Shadowrun, or possibly Android Netrunner. There are a few megacorps instead of a single government, but for the average person, life is pretty much unchanged. The justice system and social welfare are essentially nonexistent, and you might find yourself in some work camp with no warning, but the probability of this happening is low enough to be tolerable.

  36. Freddie deBoer says:

    Getting people to understand simultaneously that a) yes there were many harsh conditions in earlier eras of human existence but that also b) no life was not an unrelenting hellscape that was entirely free of enjoyment and other positive emotions and experiences is hard but necessary work. Probably important to lead to c) there is nothing especially important about right now, posterity will not remember this time as some amazing leap forward, and in a hundred years people will be making the same presentist error and assuming all of us living in 2017 must be in a constant state of pain and unhappiness because we don’t have the miracles of 2117.

    • Urstoff says:

      Aside: what happened to your education blog? I quite enjoyed reading it.

    • Randy M says:

      posterity will not remember this time as some amazing leap forward

      I’m not sure this is certain, for sufficiently long definitions of “this time.” The century from, say, 1950 to 2050, give or take 25 years either way, will surely be looked at, in technological terms, as either an inflection point or a uniquely innovative period, depending on what comes after, of course. Global connectivity, satellites, drones, genetic sequencing, etc. seem to make for a qualitatively different way of living, even if human nature remains constant.

      • Freddie deBoer says:

        I mean, we’ll see. (Not us, but future people.) But it’s worth saying that Robert J. Gordon argues quite the opposite in The Rise and Fall of American Growth, claiming that it was in fact the period from 1870 to 1940 that was the great leap forward for mankind, that progress has stalled significantly since then, and that we systematically have overrated the impact of digital technology in recent decades. Of course, there’s tons of definitional questions to argue with here. Maybe someday Scott will review that book.

        • Randy M says:

          Okay. Seems like the rate of change is, well, changing, but it’s very hard to be objective about it all.

        • Douglas Knight says:

          Scott has not addressed that book, but he has written about similar claims.

        • disciplinaryarbitrage says:

          I’ve made it about 1/4th way through Rise and Fall… and second a request for Scott’s take on it–maybe alongside The Great Stagnation (which as I understand it argues something similar).

          I readily admit that this is selfish request for someone else to get through the sheer volume of details on living standards through that time period, which basically amounts to “water, sewer, electricity, and manufactured goods are pretty great, and we got them pretty damn quickly,” and pull out some insights so I don’t have to.

      • sandoratthezoo says:

        To the extent that you believe that technology is on some kind of upward curve like a quadratic or an exponential one, note that there are no inflection points on those curves (well, besides the bottom of a parabola). There is no “elbow” in an exponential curve. The first derivative of e^x is e^x.

    • pipsterate says:

      I’m pretty unhappy about not having the miracles of 2117, actually.

  37. vV_Vv says:

    Profile of leading self-driving car researcher Anthony Levandwoski, who used to lead Waymo, co-founded self-driving truck company Otto, and now…is trying to start some kind of church relating to the Singularity?

    If memory serves me well, I think I’ve heard the claim that self-driving cars were going to be commercially available in five years for longer than five years. Then there was this legal battle between Google and Uber, centering around this Levandwoski guy, who now turns out to be a cult leader. Meanwhile, some engineers at [big company] privately told me there was no way self-driving cars were going to be ready in 5 years, and 10-15 years was a more realistic figure, which starts to look suspiciously close to the usual 15-20 years time-to-AGI prediction.

    Should I infer that self-driving cars are just extremely over-hyped and the real state of the technology is nowhere near what it is claimed by corporate PR, but likely closer to the general state of autonomous robotics, that is, not very good?

    Puerto Rico is still a disaster area. One bright spot: Google successfully restoring some communications with stratospheric Internet balloons, Tesla successfully restoring some power with portable solar arrays. Meanwhile, in the actual government…

    I’m not going to defend the Puerto Rican government, but bear in mind that actually running a country and provide people with what they need after a disaster is much more difficult than doing some flashy projects for media attention.

    Related on the new AlphaGo Zero: “I emphasize how all the mighty human edifice of Go knowledge, the joseki and tactics developed over centuries of play, the experts teaching children from an early age, was entirely discarded by AlphaGo Zero with a subsequent performance improvement. These mighty edifices of human knowledge, as I understand the Hansonian thesis, are supposed to be the bulwark against rapid gains in AI capability across multiple domains at once. I said, ‘Human intelligence is crap and our accumulated skills are crap,’ and this appears to have been borne out…[AI proved] fast enough to blow past centuries of human-style learning in 3 days”. And Robin Hanson’s response.

    Agree with Hanson here. Yudkowsky’s argument is too generic: you could also apply it to the first programmable electromechanic computers obsolescing centuries of expertise in human computing.

    • actinide meta says:

      Should I infer that self-driving cars are just extremely over-hyped and the real state of the technology is nowhere near what it is claimed by corporate PR, but likely closer to the general state of autonomous robotics, that is, not very good?

      Waymo’s cars have been driving around Mountain View for a long time, and we know from their reports to regulators that they only require human intervention every few thousand miles. So it’s pretty clear that autonomous vehicles almost work under good conditions. There’s room for a lot of disagreement about how hard it will be to strike out the italicized phrases in that sentence. Every engineer knows that the first 90% of the program takes 90% of the time, and the last 10% takes the other 90% of the time.

      Personally, I think that if our civilization wanted autonomous vehicles badly enough, we could have had them decades ago: we could have built highways solely for autonomous vehicles and equipped them with detailed markings (whether visual or buried electric cables or whatever) that allow simplistic line following control systems to navigate them. Add emergency braking sensors for safety. Design the traffic rules around the control systems. Basically, trains with wheels.

      So basically I think the question is how much “we” are going to bend to tolerate ways in which autonomous vehicles aren’t as good as humans, in order to get the ways in which they are better. If the answer is “not a nanometer”, then it will probably be a long wait. But I actually think people will decide to be a little more flexible.

      • sandoratthezoo says:

        I mean, a human intervention every few thousand miles is certainly unacceptable for the purposes of “deploying autonomous vehicles.”

        It looks to me like Waymo’s game plan is going to be to allow remote operators to takeover cars that can’t navigate themselves, and thus cut the human labor costs of taxi operations by like 90%+ and figure fuck it, the remaining 10% isn’t a big savings. Which sounds like a fairly good business move, but less impressive from a pure AI perspective.

        • faoiseam says:

          Larry is extremely hostile to tele-operation, so as long as he is actively providing oversight, I would guess that remote operation is unlikely.

          • sandoratthezoo says:

            I think you’re wrong. I mean, don’t get me wrong, the hope will be that remote operation is not often necessary, but they seem to now feel like that’s the route to not having to spend the next 4 years getting from “one intervention per 5,000 miles” to “one intervention per 250,000 miles.”

            The trick will be to make it so that the on-board AI fails safely enough that the remote operator doesn’t need to have combat pilot-level reflexes, and can instead just log in, look at the situation, and guide the car around whatever thing is happening in a leisurely manner.

      • vV_Vv says:

        Waymo’s cars have been driving around Mountain View for a long time, and we know from their reports to regulators that they only require human intervention every few thousand miles. So it’s pretty clear that autonomous vehicles almost work under good conditions.

        Yes, but do they optimize the test conditions (routes, weather, traffic, etc.) in order to look good?

        • The Nybbler says:

          Yes, but do they optimize the test conditions (routes, weather, traffic, etc.) in order to look good?

          As far as I know, they don’t run the cars in bad weather (because they won’t work), but they do run them in varied traffic conditions and routes.

    • John Schilling says:

      Should I infer that self-driving cars are just extremely over-hyped and the real state of the technology is nowhere near what it is claimed by corporate PR, but likely closer to the general state of autonomous robotics, that is, not very good?

      You should infer that about pretty much any technology that has received any significant media coverage. In the specific case of self-driving cars, we are probably only a few years out from vehicular automation that almost never needs human intervention in real-world driving, but the gap between “almost never” and “never” is likely to be A: very hard to bridge and B: intolerable while it lasts. And B reinforces A, because the best way to bridge the gap is with an enormous amount of data from beta-test experience.

      Particularly troublesome is the fact that the human assistance that is almost never required is, on that rare occasion, required immediately and with a degree of situational awareness that comes only if the driver was paying attention three seconds before their assistance was required. Finding a way around that limitation, or a profitable market for semi-autonomous cars within that limitation, would be quite helpful.

      • actinide meta says:

        Particularly troublesome is the fact that the human assistance that is almost never required is, on that rare occasion, required immediately and with a degree of situational awareness that comes only if the driver was paying attention three seconds before their assistance was required.

        Much depends on the extent to which this is true. If any significant fraction of “disconnects” really mean “there is about to be a collision and the control software can’t figure out how to avoid it” then it’s back to the drawing board until that isn’t true. And that’s certainly plausible.

        But it’s also possible that the vast majority are more like “I can’t figure out a safe and legal way [through this intersection/around this unidentified object/out into traffic that won’t yield for me] and will reasonably safely, but to the great irritation of other drivers, pull to a stop until you take over”. Frankly in that case it seems like one request for intervention per 5K miles (~6 months of average person’s driving?) wouldn’t be intolerable.

        • Evan Þ says:

          But even then, that means you need to have a licensed human driver in the car, which means you can’t use it to get home while drunk, or to send off your 8-year-old to soccer practice, or drive itself to the other end of the hiking trail to pick you up there. Unless you let a Google/Tesla/whoever employee remotely rescue the car from those situations, that obviates a lot of the advantage of driverless cars.

          • actinide meta says:

            Sure, the early market in that scenario would be (a) teleoperated taxis and/or (b) licensed drivers who prefer to work or play on their commute rather than drive. Either seems like a large enough market to easily pay for, and to John’s point collect the necessary data for, further improvements.

            It’s really popular to hate on “Level 3” autonomy as intrinsically dangerous (and the SAE definition is gibberish, so maybe that’s even technically right). But it seems plausible to me that something could be overall safer than a human driver while still getting stuck on a fairly regular basis.

          • Wrong Species says:

            A car that would let me play dumb phone games and browse social media for 99% of a daily commute is pretty good in itself. Not as great as completely driverless but still worth paying extra for.

        • John Schilling says:

          But it’s also possible that the vast majority are more like “I can’t figure out [how to proceed] and will reasonably safely, but to the great irritation of other drivers, pull to a stop until you take over”

          That, unfortunately, also falls into the “intolerable while it lasts” category, and possibly more so than the failure mode with the dead bodies. And it is likely prone to chain-reaction failures, because cars stopping without warning for no good reason (and then proceeding under the guidance of a particularly irritated human) is likely to be just the sort of thing that triggers the very same “stop and call for human assistance” behavior in the following robocars. A world in which people associate other (rich and/or geeky) people’s self-driving cars with their own prolonged gridlock, is not likely to have a regulatory environment favorable to mass adoption of self-driving cars.

      • Standing in the Shadows says:

        When I get back in the market for a new car, I’m really hoping that “autopilot” features are more widely available.

        When full in-city autonomous happens, I’ll just use robo-Uber. But when I’m driving my own car, just having lane following and adaptive cruise control is sufficient: manually pilot onto the freeway, and then just let lane following and adaptive cruise carry me to my desired exit.

        Even better if the adaptive cruise can handle actual stop-and-go traffic. Then I wont care nearly as much about commuter rush hour. Stop-and-go for an hour? Who cares? I’ll just listen to some music, catch up on my kindle queue, and chat on the phone with some of the kids.

        • John Schilling says:

          and then just let lane following and adaptive cruise carry me to my desired exit.

          Or possibly into the side of a truck at sixty miles an hour. You understand that “adaptive cruise control” requires you to remain constantly alert and undistracted and so not making phone calls, posting to SSC on your iPad, or whatever, right? If you have to be an alert and attentive driver to remain tolerably safe, then what’s the advantage to not actually driving?

          • Randy M says:

            I’d question whether its even possible to stay that alert while not driving. How often will I be lulled to sleep before the “Driver Input Needed” alarm blares me into confused wakefulness?

          • Edward Scizorhands says:

            I really doubt “having the driver take over” events will be “the human needs to figure out what is going on in the next 1.2 seconds or everyone dies.”

            It will be “the road is gone because of snow/construction/roving bands of youths, so I’m slowing down gracefully. You figure it out, please.”

            The times where things go to shit in a few seconds, you will usually be better off with a computer doing its best vs. a meatbag doing its best.

          • vV_Vv says:

            I think the reference is to the Tesla car which while in “autopilot mode” crashed into a truck killing its own “driver”.

            This was quite an issue because Tesla cars are not legally considered self-driving cars, they have adaptive cruise control but the driver is still required to keep the hands on the wheel and pay attention to the road at all times.

            But Elon Musk has played fast and loose with this, starting from calling the adaptive cruise control “autopilot mode”, and sharing youtube videos, some even made by the man who eventually died, of people using the car without their hands on the wheel. It’s like Elon Musk was implying something to the effect of “I’m not legally allowed to call it a self-driving car, but *wink* *wink*…”. Except that it really wasn’t a self-driving car and it catastrophically crashed at the first occasion it sensors had a false negative.

          • John Schilling says:

            The times where things go to shit in a few seconds, you will usually be better off with a computer doing its best vs. a meatbag doing its best.

            Explain that to the passengers of Air France 447 again? Or, as many-Vs notes, to Joshua Brown. Quite often, when things “go to shit in a few seconds”, the cleverest computer ever programmed (even by itself or other clever computers) won’t have a clue anything is wrong until everybody involved is dead. Even more often, the computer will be barely capable of recognizing, “I’m in over my head, you need to take over RIGHT NOW!”.

            Also, I’m going to propose a competition, because this community is in dire need of an insult as condescendingly derogatory as “meatbags”, to be applied at the other extreme to idiot AIs and the idiots who worship them. Ideas, anyone?

          • Controls Freak says:

            Jumble o’ Tiny Cords and their adoring fans, the Resistorheads?

            Light-up Boxes and the Dead Bulbs?

            Rube Tubes and the Goldbergs?

            Hollow Desks and the Turkmasters?

            Space Heaters?

          • The Nybbler says:

            this community is in dire need of an insult as condescendingly derogatory as “meatbags”, to be applied at the other extreme to idiot AIs and the idiots who worship them. Ideas, anyone?

            Wireheads? If you don’t like the conflict with the Niven meaning, perhaps wirenuts?

          • Lillian says:

            The classic counterpart of meatbag is tincan, which i find is a classic for good reason. If you really want something new to use specifically for deficient AIs and their overly enthusiastic proponents may i suggest dumbots and dumbotters?

          • Nabil ad Dajjal says:

            Personally I’m a big fan of “toaster.”

          • Edward Scizorhands says:

            Fine, I’m going back to “ugly bag of mostly water.”

          • John Schilling says:

            “Toaster” is a good choice if the Battlestar Galactica reboot is part of the cultural canon. That may be a reasonable assumption here, so I may see how it works.

          • Edward Scizorhands says:

            Even more often, the computer will be barely capable of recognizing, “I’m in over my head, you need to take over RIGHT NOW!”.

            Right. Like I said, there aren’t going to be many situations where the computer realizes a) it is in trouble, and b) a human can then react better than the computer can. By definition, there are over 9 million drivers in the bottom 5% of driving ability on the road right now. And even humans in the top 5% won’t be able to hit the brakes as fast as a computer can.

            So for a while, self-driving cars will need humans to be conscious, as in “not asleep” and “not drunk off their keister” but it’ll be okay if they are posting anti-human memes onto SSC.

          • Nornagest says:

            “Bitfucker” comes to mind.

          • John Schilling says:

            So for a while, self-driving cars will need humans to be conscious, as in “not asleep” and “not drunk off their keister” but it’ll be okay if they are posting anti-human memes onto SSC.

            A person whose only virtue is that they are conscious ans sober, requires 5-6 seconds to take effective control of a vehicle. If the truck in front of you just dropped something that might be a cardboard box and might be a crate full of anvils, 5-6 seconds is far too long to get a competent decision on whether you should swerve into the next lane and hope the traffic already there will all get out of your way. And that’s before we get to the truck that somehow looks just like the empty sky so the toaster won’t even know to ask for help.

          • Standing in the Shadows says:

            might be a crate full of anvils

            This isn’t India. If the truck in front of me is full of crates of anvils that are even remotely able to be accidentally dropped on the road, then the driver is about to lose her CDL, and the driver’s employer, the owner of the truck, and the loadmaster are all about to lose their bonds, and their insurance, and 6 insurance companies are about to be very very pissed off.

            If I am passing behind a truck and a load of anvils is suddenly abruptly dropped on the road right in front of me, my life is not going to be saved by a selfdriving AI suddenly playing trolley problem, its not going to be saved by a adaptive cruisecontrol hitting the brakes, and it’s not going to be saved by my awake and attending meat brain spending 1700 miliiseconds to hit the brakes. It’s going to be saved by the design engineers in Tokyo and Detroit who years ago engineered the crash crumple process of my car, and who engineered the seatbelts and airbags.

        • baconbacon says:

          “the human needs to figure out what is going on in the next 1.2 seconds or everyone dies.”

          Most of these events can be avoided by increasing following distance/driving at the correct speed/not being drunk. In almost every situation ‘slow down’ is a good choice.

        • Deiseach says:

          I think if there are mass-market self-driving cars, they will be for highways and autobahns and the like: long stretches of flat roads where the speed, lanes, traffic flow, etc is fixed and directed. For city driving the human driver will have to take over because it’s too unpredictable and stop-start and the streets are narrow and crowded and you have buses, cyclists, pedestrians and all the rest of it. So no “relax and listen to music as my self-driving car navigates the jams on the road into work”, unfortunately. It’s hard enough to work out how to regulate all the traffic lights, regulating thousands of cars to go and stop and slow down and turn off here in a city? No chance!

          • It’s hard enough to work out how to regulate all the traffic lights, regulating thousands of cars to go and stop and slow down and turn off here in a city? No chance!

            As I understand it, the previous version of this was the intelligent highway, with one entity controlling all that, and people eventually concluded that it wasn’t workable. The current version is decentralized, with each car controlling itself just as it’s done now–but with a computer instead of a human making the decisions.

          • Edward Scizorhands says:

            “Wasn’t workable” in that there was no way to bootstrap there, or that it could never work at all?

            After a significant number of cars on the road are intelligent, you can start having intelligent traffic lights, that dynamically direct cars.

          • Nornagest says:

            The problem with the old intelligent highway concept is sensors, basically. You could build some simple lane-following logic into cars with 1950s technology — just lay down a conductive strip in the middle of the road and use analog control systems to get them to follow it. Add centralized systems to manage cars’ positions relative to each other and we’re talking 1980s technology, but still nothing too complicated in the 95% case.

            But what happens when the wind blows a bunch of tree branches down, or one of those mysterious empty trash cans appears on the highway however they get there, or a robot semi-trailer blows out one of its tires, spraying a hundred pounds of rubber shrapnel all over the road? The lane-following system isn’t going to see the debris, and it won’t be tagged as a car, so the centralized system won’t see it either. It’ll only know there is an obstacle when the cars it’s managing start to crash, which will create more debris, and it’ll be totally unable to clear it or route around it. It’ll be a stupendous fustercluck until human workers come out to clear the road, haul off the wrecks caused by the first wave of interactions, and restart everything. That’ll take hours, and it’s not a rare enough kind of situation that you can afford hours of downtime.

            To avoid that, you need sensors — and, more importantly, logic — capable of the same kind of hazard-avoidance tasks that human drivers do. That’s cutting-edge stuff, and once you have them, there’s no point in doing it at the highway level because all the 1950s and 1980s infrastructure has just become obsolete.

  38. analytic_wheelbarrow says:

    On how police unions can either fight or reinforce corruption: “The things cop unions do that reformers don’t like – reflexively defend all officers in all situations, fund legal defenses and media campaigns…those are all felt, by cops, as safeguards against police corruption”.

    No comments on this? I thought this was the best link of the bunch! Especially since it took something I “knew” and turned it upside-down.

    • Mengsk says:

      Agreed– this was really interesting, and perhaps not done justice by the quote. The TLDR is that, by reflexively defending officers, unions make retribution within the police force more difficult. The nature of police work makes it pretty easy for cops to punish whistleblower types. For example, if an officer exposes criminal activities within the force, they could easily end up being put into a dangerous situation without adequate backup.

      Similarly, when the public opinion favors a crackdown on X, it’s possible for corrupt actors within the institution to use the demand for the crackdown as an excuse to punish or marginalize their enemies within the institution while appearing to be responding to the public’s demands.

      • The Nybbler says:

        Similarly, when the public opinion favors a crackdown on X, it’s possible for corrupt actors within the institution to use the demand for the crackdown as an excuse to punish or marginalize their enemies within the institution while appearing to be responding to the public’s demands.

        Of course they do. That’s why if you want an honest police force, you have to take the entire thing down (including the court system as well, which treats the cop’s word as gospel. Heck, here in NJ an accusation from a cop is prima facie proof of guilt in a traffic case, and you don’t even have a right to cross examine), disqualify anyone who has ever been a part of one, and start from scratch with some sort of different founding principles. But we’ll never get that.

  39. TheApiary says:

    My question about psychiatric diagnoses: what exactly is the point of worrying about them, aside from that we have to call it something for the insurance company, given that there is so much overlap between treatment options for the various diagnoses? What exactly do we gain from specifying that this is bipolar II instead of MDD + GAD aside from getting a good guess of what medications to try first? If it’s only that use, is there any sense in-rediagnosing people once we’ve tried a few medications? That is, if someone was diagnosed with MDD and GAD, didn’t improve on SSRIs and is now doing great on Latuda (or whatever), why should we care if she actually has bipolar II, and in what sense is that a meaningful statement?

    • dhominis says:

      I’m assuming that by MDD + GAD vs. bipolar II you mean that the hypothetical patients has clear-cut depressive episodes and also has times when she’s unusually agitated/insomniac/hyper-focusing on specific things (is she suddenly spending twelve hours a day at work because she’s anxious or because she’s hypomanic?)/irritable.

      Identifying this hypothetical patient as experiencing something like mania can be helpful in and of itself. In some people who would probably be best diagnosed with bipolar II, SSRI use can lead to (hypo)mania. If she ever goes off of Latuda and onto an SSRI, she should be monitored for signs of mania or worsening hypomania. More generally, saying she has bipolar II tells you something about her potential to develop worse bipolar symptoms in the future.

      In terms of nonpharmaceutical interventions, identifying her problems as GAD and not hypomania gives you a better idea of what to do for psychotherapy, too.

      Agreed that specific diagnostic labels can be less important than some psychs seem to treat them as. Anecdotally–when I was heavily involved in the psych system, there were a few multi-month debates over questions such as “should we diagnose him with schizoaffective (depressive subtype), schizophrenia and major depressive disorder, or schizophrenia and depression NOS?” Eventually my doctors settled on psychosis NOS and depression NOS, along with several others–this was based on my having recovered completely (and therefore never having been schizophrenic despite several consecutive years of hallucinations+delusions+total flat affect/anhedonia+disorganization w/ significant functional impairment…), not on the symptoms. Fun stuff.

      I’ve wondered about moving from diagnoses to symptom descriptions: not “does this patient have GAD+MDD or BPII?” but “this patient has [for example] discrete periods of time with dysphoric mood/guilt/anhedonia/decreased goal-directed activity, discrete periods of time with agitation/euphoric mood/increased goal-directed activity, diffuse and general anxiety that isn’t limited to discrete periods of time.” This would be complicated, obviously, and symptoms themselves are subjective, and I’m sure there are quite a few problems I haven’t considered… but it’s interesting to think about. At least you wouldn’t have patients with no symptoms in common and the same psych dx or patients with similar symptoms but totally different dxes.

      • TheApiary says:

        Oh the implications for therapy are interesting, I hadn’t thought of that. Diagnoses work reasonably to give a good first try medication, but given that there are so many bipolar patients who take SSRIS with no problem, and apparently unipolar people who respond better to mood stabilizers or AAPs, I don’t think they’re very helpful after the first couple things you try do or don’t work.

        I’ve also thought about the symptom-based descriptions. They definitely would be more accurate and precise in a lot of ways. I imagine that they’re much harder to use in big clinical settings, because there would be too much to read for a doctor to scan in one second at the top of a chart. But I wonder if they could be more useful as we move into more computer diagnosis. It seems like if, instead of just “bipolar disorder,” the computer had a specific list of individual symptoms and then knew which drugs did or didn’t work for that patient, it would have an easier time learning something useful about how to match drugs to symptoms. There are already a lot of studies like that, along the lines of “depressed patients with x symptom respond more to drug y than other depressed patients do” but which things they try seems kind of haphazard. If that were standard operating procedure, imagine all the patterns we haven’t thought to look for

        • dhominis says:

          Good point about bipolar people who do well on SSRIs. I’d still argue that it’s useful to know to keep an eye out for worsening hypomania/mania (when sleep-deprived, under stress, on stimulants, et cetera).

          There has to be some way to get around the longwindedness–if you limit what’s at the top of the chart to more general symptoms/categories (say, “major depression, possible hypomania, diffuse anxiety, current mood normal” for this hypothetical patient), that might not be too bad.

          Training a diagnostic program based on symptoms instead of on reasonably fallible diagnoses based on symptoms does seem like it could provide important insights. This discussion reminds me of that mental disorders as networks post–basing the conceptualization of mental illness more on the specific characteristics than on the aggregate diagnosis.

          • TheApiary says:

            There are also people who’ve never been manic before who get manic on SSRIs, and I’ve read a few people arguing that that means they “really” have bipolar. But that seems very unhelpful if there’s no way you could know they have bipolar disorder before you tried SSRIs, and mood stabilizers and antipsychotics are what people try for depression if antidepressants don’t work anyway.

            Yes! I was thinking about that post and trying to remember enough of the title to find it. I guess at that point we could pick a few of the symptoms, like you’re saying, to put on the top of a chart and keep the rest in the computer to network. I wonder how that would be different– I wouldn’t be surprised if doctors accidentally treated “bipolar II” differently from “depressed mood, irritability, agitation”

  40. maybe_slytherin says:

    The old “everything looks like Arabic” post is gold.

    Key quotes:

    Barack Obama is secretly an angelfish

    In Arabic, the curves in this river mean “I was formed by the random action of erosion, and my spelling out this legible sentence is only a coincidence”

  41. Conrad Honcho says:

    President Xi’s response to learning people thought he resembled Winnie the Pooh: “Oh, bother.”

  42. Kevin C. says:

    Put me down as another one who mostly agrees with Hanson’s linked criticism of Yudkowsky’s argument. But, I must note that I don’t find Hanson’s overall view of the likely future much more plausible. Is anyone else arguing the view that the likely scenario is that, for the forseeable future, we’ll have neither AGI nor “ems”, but instead futher progress along the lines of “narrow” AI like AlphaGo and self-driving cars, which are “superhuman” at their particular tasks and only those tasks? Where said machines “eat” large chunks, and eventually the majority, of the activities through which human beings contribute to the economy, as per, for example, this Kevin Drum article*, but no “foom” to the AGI-driven Singularity nor Malthusian “ems” economy nor Trekian post-scarcity utopia, but something more like Cowen’s Average is Over scenario?

    *Also interesting was Rod Dreher’s recent engagement with that article, and what the “conservative” response should be to mass technological unemployment, but discussion of that is probably best left to the next open thread.

    • Le Maistre Chat says:

      I don’t believe in the Singularity or that plausible post-scarcity technology would make a utopia, but I don’t know what the Average is Over scenario is.

      • sandoratthezoo says:

        Average is Over means basically life gets considerably better for the top n% of humanity (n < let's say 40) and much worse for the lower 100-n% of humanity. So it would be a scenario in which ultra-skilled high-IQ work gets a giant productivity multiplier on it and is very well-rewarded, while low-skill work gets automated away. And/or one in which capital becomes much more productive, leading to people who have capital doing very well and those who don't get stuck.

      • Kevin C. says:

        I don’t know what the Average is Over scenario is.

        Something like the top ten to fifteen percent of the population becomes highly productive and fabulously wealthy because they have the skills/talents that “complement” automation and are still the domain of human beings; some few percent more make a living providing artisinal personal services to those folks, or otherwise serving as a “toady class” to the elite, and often finding their way into the lavish, personal-security-protected gated communities into which those elites will retreat; and the remaining ~80% mostly live in Mexican-style favelas eating beans*, living off meager welfare/UBI/redistribution supplemented by Amazon Turk-style make-work “for pennies”, but they’ll have plenty of “virtual experiences” and other cheap electronic entertainments, and if that, population aging, and widespread provision of psychiatric drugs aren’t enough to pacify the underclass, improvements in surveillance, drone, and police technologies will handily suppress any outbreaks of unrest.

        *Not joking:

        Poor Americans, writes Mr. Cowen, will have to “reshape their tastes” and live more like Mexicans. “Don’t scoff at the beans,” he says. “With an income above the national average, I receive more pleasure from the beans, which I cook with freshly ground cumin and rehydrated, pureed chilies. Good tacos and quesadillas and tamales are cheap too, and that is one reason why they are eaten so frequently in low-income countries.”

        (from Philip Delves Broughton’s book review of Average Is Over, in the Wall Street Journal, Oct. 2, 2013]

        • Le Maistre Chat says:

          Oh, yeah, that makes perfect sense to me. The Have(-capital)s are far wealthier than today, while those who Have Not because their jobs have been automated away live off welfare in high-crime areas.
          I’m agnostic on how easy revolutions will be to suppress.

          • Watchman says:

            But every historical comparison (and we have had a lot of automations over time, with the industrial revolution perhaps being the most obvious, but the invention of the plough perhaps the most important) suggests that this does not happen. Whilst rapid automation creates short-term unemployment, it does not create a permanent class of unemployed people, never mind concentrates the wealth to a small minority – yes, the 1% are fantastically rich, but compared to my great-great-great grandfather, a market gardener in Slough (yes, that’s a place – pronounced Slouw…) I am also fantastically rich, and we probably occupy similiar positions in terms of average income and social status. And to draw on a recently discussed example, no-one globally is as comparatively rich as the late lamented Ogdai Khan was in his day because there is a lot more wealth nowadays with which to be fantastically rich.

            Even now, when the internet has revolutionised industries and done away with vast swathes of jobs that used to exist, we do not have massive unemployment as a result. Human’s adapt – it’s why we’re the dominant species. As as long as humans (or AIs) have needs that we can meet, we will find ways to do this that match our skill sets.

            The risk here is political actors intervening on the basis (which seems popular amongst left-wing/liberal media commentators) that automation will destroy jobs and not create others, an all else being equal assertion that seems highly unlikely. The actions set up to mitigate against this probable non-event, being pretty much rehashed modernist approaches to social safety nets applied on top of exisitng systems, would create the conditions to allow the development of the have-not class, supported by the state.

          • The Nybbler says:

            Whilst rapid automation creates short-term unemployment, it does not create a permanent class of unemployed people,

            Yet we already have such a class. There’s pressure against it, mainly that it can be cheaper to use people than automation (cheap Chinese labor probably set automation in electronics manufacture back a few years), but things like the minimum wage and benefits and safety regs push back on that also. Why wouldn’t it get worse as automation gets better?

          • bbartlog says:

            @Watchman : previous automations may not have created a permanent class of unemployed people in part because the newly unemployable slowly died. Surely you are familiar with the Swing Riots? Have you read Jack London’s ‘People of the Abyss’?
            I don’t particularly think this is an argument against automation, because given the near-Malthusian constraints of the time, many or maybe even more people would have died in the absence of the improved productivity that was being rolled out. It would just have been a slightly different and somewhat larger set of people. But to claim that historically there has been no ‘permanent class of unemployed people’ flies in the face of the first-hand accounts of the time. Or at any rate, if such a class was not permanent it was not for the most part because they pulled themselves up by their bootstraps, but because they died.

          • John Schilling says:

            But every historical comparison (…) suggests that this does not happen.

            Every historical comparison involves automating away jobs that were well below the intellectual ability of average humans, but which nonetheless demanded the attention of millions of average humans because they absolutely had to be done. This freed up average humans to do other, better things. The present round of automation is doing away with some jobs that are a good match for the intellectual ability of average humans, freeing up average humans to do, well, that’s not exactly clear because we aren’t automating away all of those jobs, but history is probably not a good guide as to how this will turn out.

          • baconbacon says:

            @ John Schilling

            I don’t think this is at all true. Being a baker now might not be a particularly mentally taxing job, but that is with the system currently in place. A century ago your flower wasn’t delivered with machine precision mill, your bakery wasn’t climate controlled and your yeast wasn’t grown in a sterilized vat. In addition if you were semi-literate or worse that meant memorizing all of your recipes and/or developing your own rudimentary language for recording.

            As evidence to support this most artisans had apprenticeships that lasted multiple years of more than full time work before moving on to some intermediate stage before becoming a stand alone artisan. Virtually all of these jobs (as a percent of the economy) have been displaced by automation.

          • Nabil ad Dajjal says:

            @baconbacon,

            This is possibly the pettiest nit I’ve ever picked, but it’s flour and not flower.

            In more substantial terms, baking is a weird example to use because it’s pretty bifurcated. A baker along the lines of a pastry chef is a skilled tradesman and that takes a fair amount of mental ability: they’re not rocket surgeons but it’s not a job that anybody off the street could do. But the vast majority of baked goods are produced in factories where the workers could probably be replaced by machines pretty easily.

            The long apprenticeships you’re talking about basically still exist for the tradesman type bakers but the equivalent of assembly-line type bakers is probably closer to peasant farmers. If we get rid of all of the simple manual labor that won’t magically turn those people into skilled workers, just reduce their already limited options.

          • baconbacon says:

            @ Nabil ad Dajjal

            I’m usually posting while giving my kids lunch etc, I am always going to make ‘mistakes’ like flour/flower, personal/personnel etc.

            The point was that most of these jobs from centuries past have already be eliminated through automation. Millers have been put out of work en masse, there used to be tens of thousands (possibly hundreds of thousands) of small mills scattered across the US, and you had better believe that running them took a large skill and knowledge base. There were high intellect jobs all the way through the production structure, the guys that owned the teams of horses/oxen that pulled the grain to and from the mills had to be amateur veterinarians, the blacksmiths that shod them were experts at a complicated trade etc, etc. These jobs have been heavily displaced as well as the lower intelligence farm laborer stereotypes.

          • The Nybbler says:

            The comparison isn’t baking, it’s ditchdigging. For instance, the Delaware and Raritan Canal — one hole 44 miles long, 75 feet wide, and 8 feet deep, and a smaller hole 22 miles long, 60 feet wide, and 6 feet deep — was dug in 4 years, by hand, employing roughly 3000 people (perhaps not all at once, as many died and presumably were replaced). 6.7 million cubic yards if I’ve done the math right.

            A big excavator can do very roughly 600 cubic yards per hour. Which works out to one excavator team (including dump trucks, etc) being able to dig the whole canal in 4.5 years, based on 8 hours of earthmoving a day and a 6 day week (probably what the laborers were doing). Of course it isn’t that easy; you can’t dig in parts of the winter (by hand or by excavator) and there’s lots of other stuff besides digging the hole. Still, replacing 3000 laborers with a couple of excavators (and trucks and other equipment) and a few dozen people to run them is a huge reduction in labor use.

          • baconbacon says:

            @ Nybbler

            The quote was

            Every historical comparison involves automating away jobs that were well below the intellectual ability of average humans

            The fact that ditch digging has also been replaced is irrelevant. High skill/high intellect jobs have been displaced in the past, which directly counters Schilling’s claim.

          • Watchman says:

            @bbartlog

            Whilst it is not my period of study, I’m well aware of the industrial revolution (I live near its birthplace, and grew up in a landscape defined by it, and by the revolutions that ended the mass-industrial phase). There were indeed always poor people after automation of things like weaving and milling. But there were also the same poor people before automation – just check records of any seventeenth or eighteenth century English (or I suspect, New England) parish to find the records of the parish delivering its charity. I don’t believe the continued existence of this group of poor people proves automation was the problem – it either proves their is always going to be some poor people (and not necessarily always the same people – the records show churn, not constant poverty) or that the economy was not working to full efficiency (sport the free marketeer…). As to the argument that the specifically unemployed slowly died, in Britain at least in the industrial revolution we know that isn’t the case – they didn’t remain unemployed in the countryside but were recruited to work in the cities because the industrial revolution, despite making production more efficient, actually needed labour to meet demands for products that previously might not have been sought such as pen nibs or affordable buttons.

            And that is the key point that this misses – every phase of automation in history has made us wealthier, and that wealth has created demand for more consumer goods/services, and that has created employment (often on better terms and conditions as well). Why this should change with another bout of automation is not clear to me.

            @ The Nybbler

            I tend to believe the permanent class we have now is mainly created by peverse incentives and by bad life choices (unfortunately these are often the parents’ choices…). I do see this class being increased if the sort of politicised reaction to automation I mention above is introduced to be honest, but otherwise I think it is a combination of the unavoidable poor (the losers in the churn, at least supported properly nowadays hopefully) and those taking the easy, if hardly luxurious, option of living on the state.

            @ John Schilling

            Intellectual ability is an odd one. My son can use a remote control to turn on a television (and then to lose the picture as he presses a random button – but he knows one button to press…), and he knows that swiping a tablet does things – he’s ten months old and apparently (to the disappointment of my paternal pride) pretty normal in doing this. How do we measure that sort of background knowledge in this equation. We aren’t getting in The Nybller’s ditch diggers to compete against automation, but people who are mostly familiar with modern technology and interfaces (and my 70 odd year old parents use tablets, so age is not really an argument here either).

            Also, any technological advance with an interface that is notably beyond the average intellectual attainment of society is unlikely to be an advance that is adopted (I might argue UNIX as an example of this) because it makes using it more expensive than the economical alternative, and is likely to lose out as a result (mental note – do some digging on why Betamax lost out…). So whilst the process might be beyond the average person’s understanding, the interfaces and tasks a job involves would not be. It’s not as if the mill workers of the 1800s knew how to fix the machines they used – they knew how to use them and who to call if the machine broke.

          • On the subject of what happens to people who are unemployed because the jobs they used to do no longer exist …

            There are at least two ways in which government intervention in the market unnecessarily increases the problem. One is the minimum wage, which means that if your productivity is below that level you cannot be legally employed. That’s particularly a problem for people entering the labor market, so means that they may never be able to take the first step towards better paid employment.

            The other is professional regulation, licensing and similar restrictions, which raise the cost of entering professions many of which, such as barbering or hair braiding, do not require high levels of professional training.

            That’s a good deal of the reason that the one charity I regularly give money to is the Institute for Justice, a libertarian public interest law firm that fights such regulations in the courts, sometimes successfully.

            There is a third way in which government intervention increases the problem, but that involves a real tradeoff. The more generous the provision for unemployed people, whether through unemployment insurance or welfare, the more people will choose not to bear the costs of employment. On the other hand, the less generous the provision, the worse off people who are unemployed through misfortune will be.

            All of this raises a question that I don’t think I have seen much discussed. Suppose technological progress makes it possible to support anyone who doesn’t have a job at a moderate but tolerable level, say the equivalent of twenty or thirty thousand dollars a year per capita. Are we better off providing that to anyone who wants it and producing a society where a sizable part of the population is permanently unemployed, contributes nothing, entertains itself with video games, social status competitions, television, and the like, or better off if we somehow arrange things so that almost everyone is working, with welfare only for those who really cannot. For my current purposes, I will assume that the second scenario results in the person who would have had a free ride at thirty thousand having a job at something somewhat over that–but not enough over that to compensate him for the leisure and effort the job costs him.

            I should add that this is one of several issues implicitly discussed in the science fiction series I commented on recently, starting with Torchship, although it only becomes important later in the series.

          • Nancy Lebovitz says:

            “a sizable part of the population is permanently unemployed, contributes nothing, entertains itself with video games, social status competitions, television, and the like”

            To some extent, they’re company for each other, and possibly for employed people as well.

            Depending on how much they’re given, some of them might even be organizing events without getting paid for it.

          • They’re also consuming — somone’s got o.

          • it does not create a permanent class of unemployed people,

            Yet we already have such a class. There’s pressure against it, mainly that it can be cheaper to use people than automation (cheap Chinese labor probably set automation in electronics manufacture back a few years), but things like the minimum wage and benefits and safety regs push back on that also. Why wouldn’t it get worse as automation gets better?

            Firstly “level of unemployment” does not imply “class of permanently unemploeyd people”. Those are two different things. Unemployment has its uses — it holds down wages and thereby inflation…although , for some reason, I always get pearl-clutching when I point that out.

            Secondly. automation can barely be said to cause anything outside some framework of ownership, redistribution, etc.

          • Edward Scizorhands says:

            is permanently unemployed, contributes nothing, entertains itself with video games, social status competitions, television, and the like

            Don’t forget attending Klan or AntiFa rallies to beat the shit out of people they don’t like. Or posting on twitter all day to build up mobs to get the people who do the useful jobs fired for expressing something they disagree with.

          • Jack says:

            @Edward Scizorhands
            Is there evidence that unemployed people are more likely to attend Klan or Antifa rallies, or to use Twitter to get people fired?

            And if the answer is yes, a follow-up: might there be a difference in the kind of experience of social alienation that (I hypothesise) can turn unemployment into violence, versus how people would feel about “unemployment” in a world where not doing market work is common and acceptable?

          • Edward Scizorhands says:

            There is evidence that the people at the Charlottesville or Antifa rallies or the people going around ginning up Twitter mobs are not gainfully employed or have trust funds backing them. I realize that’s B->A instead of A->B.

            Is it theoretically possible that the reason these people do these acts is just because society looks down upon them? Yeah, I guess it’s theoretically possible. Just like it was theoretically possible that the reason that people using food stamps committed more crime was because society looked down on them as so we should get them an EBT card indistinguishable from the credit/debit card a middle class person uses.

            All in all, I’m very suspicious of any explanation that we can solve social problems via demanding that people’s social status be raised. Because 1) you can’t control that; people are judgmental jerks and simply dictating through law that the old ways of judgment are now verboten means that they will find new ways of doing the judgment, and 2) it’s often the underlying problem that’s causing both the social distaste as well as the antisocial behavior, and 3) it’s unfalsifiable. I can come up with a thousand third parties to blame for my behavior.

        • Is Cowen holding typical US levels of welfare constant? A lot of people who make dystopian predictions about UBI are not making a prediction for the sake of making a prediciton, they are trying to nudge things in a more leftish direction. And in countries that already have gernous welfare systems, people worry a lot less about robots doing all the work.

          Hands up anyone who would want to be part of a libertopia where they are 80% likely to be unable to sell their labour?

          • Nabil ad Dajjal says:

            Hands up anyone who would want to be part of a libertopia where they are 80% likely to be unable to sell their labour?

            You’re only 80% likely from behind a veil of ignorance. In reality, you already know roughly where in the distribution you fall and can place solid odds on where your kids will fall.

            Pretty much everyone on SSC is in the 20% of suckers who would have to keep working in order to support the lifestyles of the 80%. That sounds like one hell of a raw deal.

            I would never consent to that unless the franchise was restricted solely to the employed. Otherwise you’re signing up to be the one sheep voting against four wolves on what to have for dinner.

          • Standing in the Shadows says:

            Pretty much everyone on SSC is in the 20% of suckers who would have to keep working in order to support the lifestyles of the 80%. That sounds like one hell of a raw deal.

            In my own social media places, where I care to engage about this, whenever UBI comes up, I point out that 100% Mind-managed automation is not going to happen, especially not at first, and so there will still need to be humans working at non-automatiable and diffcult jobs that will require hard work and talent, and how do they propose to keep those jobs staffed?

            Because the people who can do that work are not going to do it for a 98% tax rate and the sneers of the riders of the purple wage.

            When I’m particularly grumpy, I frame it as “if you want to live like an Eloi, I want to dine like a Morlock”.

          • Nornagest says:

            Well, that puts a new spin on “eat the rich”.

          • Indeed. Disenfranchising the 80% is not enough, you need to find a way to stop them rising up. I can see how that can be done with a repressive state apparatus, although that’s not libertarianism, and I can see how that can be done be keeping the m comfortable and distracted, but that’s reinventing UBI.

            Basically

            A. extreme inequality

            B. No coercion

            C. No taxation

            ….choose two.

          • johansenindustries says:

            “Indeed. Disenfranchising the 80% is not enough, you need to find a way to stop them rising up.”

            It seems to me that in the case with some huge inequality without UBI and that. That the 20% can quite happily watch them rise up. They merely have to withdraw their consent to be taxed and for each to use some of him extreme wealth to keep him and his safe.

          • Edward Scizorhands says:

            This is another reason I support wage subsidy over basic income. It’s a lot easier to get the necessary 20% on board if the other 80% are working.

          • Nabil ad Dajjal says:

            Disenfranchising the 80% is not enough, you need to find a way to stop them rising up.

            If 80% of people are extracting UBI from the 20% who still work, then they’re the ones on top under that system. Honestly it sounds more like “cracking down” to me.

            After all, you don’t pay your employer for the right to work: even if you love your job, they still have to pay you to do it. Getting paid out of someone else’s earnings has traditionally been a privilege of the nobility and the clergy.

            To me this has always been the most objectionable part of UBI: it’s taking from the ant and giving to the grasshopper.

            I work hard to do something valuable for society that most people are simply incapable of doing. And I wasn’t born into my job, I sacrificed a lot of other opportunities in order to get to this point. I didn’t have to do any of that: I could easily have been less ambitious and taken an easier path. A UBI is society’s way of saying “fuck you” to the able and the ambitious.

          • @johansenindustries

            Private armies are coercion.

            @EdwardScissorHands

            What incentive does wage subsidy give to employers?

          • Edward Scizorhands says:

            I don’t think I understand your question but I’ll try.

            It gives more employers incentive to hire for more positions. Maybe it wasn’t worth hiring someone to be a greeter at the front door, but with the wage subsidy pushing the market-clearing-wage (for the employer) down to $5, it pays off in terms of loss prevention and customer experience.

            It makes some other businesses possible that wouldn’t be without it, pushing wages up for other people.

            I don’t believe people* are unemployable. I believe their market wages may be too low to be a) worth the work or b) above the minimum wage. Wage subsidy brings them back into the marketplace, and every new job that gets created pushes up wages for everyone else.

            (* I mean most people. Terri Schiavo was unemployable, but in the modern electronic world it takes a serious combination of physical and mental impairment to not be able to provide economic value to someone else.)

          • The Nybbler says:

            (* I mean most people. Terri Schiavo was unemployable, but in the modern electronic world it takes a serious combination of physical and mental impairment to not be able to provide economic value to someone else.)

            Unable, yes. Unwilling, not so much.

          • Edward Scizorhands says:

            Yes, but that’s a problem we already have right now. Wage subsidy makes that problem no worse and potentially makes it better.

        • hollyluja says:

          Isn’t this “The Diamond Age” scenario?

    • Wrong Species says:

      You’re comparing apples and oranges. Cowen looks at the next 20-30 years. I’m not sure about Yudowski specifically but Bostrom seems to think AGI will happen sometime in the latter half of this century. Hanson predicts EMs in a couple hundred years. If by forseeable future you mean closer to Cowens timeline then we probably won’t see AGI. On a longer time scale, it would be shocking if we don’t create beings more intelligent than we are now.

  43. mobile says:

    The flip side of cities racing to the bottom to attract investment from Amazon is for cities to treat existing large capital investments as their personal ATMs, as a company cannot easily move an existing facility to a neighboring state or city. In bad times or ambitious times, the temptation is high to balance the city budget on the backs of their most immobile residents with the deepest pockets. In the long run, local companies that are still successful tend to invest someplace new. There are significant economies of scale and other benefits for, say, GM to build all its cars (for the US market) within 20 miles of Flint. Instead, GM has manufacturing facilities in 10+ different states, and this can be seen as diversification against tax and regulatory risk.

    Amazon is not the monopoly. The monopoly here is the local government that has secured a large capital investment.

  44. Nornagest says:

    Independent: “each day, a cruise ship emits as much particulate matter as a million cars. They say that “30 cruise ships pollute as much as all the cars in the UK”, but I wonder if that’s true

    It’s probably true, but very very cherry-picked. Four-stroke gas engines, the type you find in cars, emit very little particulate matter. Two-stroke engines like you’d find on a lawnmower or a weed-whacker emit more, and so do the diesels in trucks, but residential fireplaces, wildfires, dust, and the coal industry (mining and burning) account for more than that. If you’re near the ocean, sea salt is a major source.

    For CO2 emissions or for many other pollutants, the cars would win by a mile.

  45. John Schilling says:

    The only time you’re allowed to sign up for Obamacare insurance policies this year is between November 1 and December 15 (slightly longer in some states).

    Between November 1 and December 15, OR whenever you have a “qualifying life event”. The list of which includes things like losing a job, getting a better-paying job, getting married, getting divorced, moving…

    Finding clever ways to game the system via carefully scheduled and/or opportunistic QLEs is left as an exercise to the student. But if you carelessly wake up on December 16 without insurance, there are options.

  46. sov says:

    It’s even wackier: if you removed the top 10-15 most polluting ships, it would be equivalent to removing every car on the road. Most ships burn bunker fuel, and because they aren’t really tied to any particular set of environmental rules in international waters they’re free to select the most dilemma-defecting choices. There’s (particularly in the Nordic countries) been more of a push to put lithium ion batteries in ships. Basically, they’re already cost effective amortized over 3-5 years but government subsidies bring that down enough to be worth the risk.

    • antimule says:

      Ship powered by lithium ion? How long can engine last on that?

      • Ketil says:

        “Harmony of the Seas” apparently can generate up to 100MW in total, and a bit of googling reveals that cruise ships typically can consume 100-200 tons of fuel per day. One TOE is roughly 10000 kWh of energy, or about 100 of the largest Tesla batteries, each weighing about 600kg. So you’d need something like 10-20K batteries weighing 6-12 000 tons (a few percent of total weight, but sixty times the weight of the fuel), per day of electric power.

        Pure electric doesn’t sound like a great idea (And where the hell would you charge a thing like that anyway? Or get enough lithium for more than a couple of ships?), but possibly enough battery for a day or so in port might be feasible. A few (six for Harmony) thousand inhabitants probably consume 10-20kWh/day each, i.e. maybe 20 tons of oil, or 1000 tons of lithium. But you could also use a LNG gas turbine for that – or just put a reactor in there, of course, and solve all your problems in one go.

    • J says:

      The Lithium-ion claim sounds pretty crazy, but to your point about bunker oil, I’ve read that ships will switch to cleaner burning fuel as they get close to port. Also, I think bunker oil is basically the crudest of refined petroleum products, so I’m curious what would happen to it if it wasn’t going to the ships. Would it get refined (potentially in a dirty process), burned in a land-based plant, something else?

      • Nornagest says:

        It’s the heaviest petroleum product. Petroleum refining works by horfing a bunch of crude into a big old steel-cased fractional distillation column the size of a 747, which separates the light oils from the heavy ones by evaporating them off and letting them condense higher in the column where it’s less warm. A bunch of pipes stick out the sides of that column to tap off distillation products — imagine spigots pounded into a wine barrel — and the higher they are, the lighter and less viscous the stuff they suck out is. Near the top, you get gasoline. Near the bottom, diesel. Bunker oil is the dense, sludgey, nasty stuff that collects at the bottom, like the wine barrel’s dregs.

        You can’t refine it any more because it’s already refined. You could potentially crack it into lighter oils with the right chemical process, though, or process out some of the impurities. But only if there was a market for it.

  47. youzicha says:

    More Section 58 news this week: How the U.K. Prosecuted a Student on Terrorism Charges for Downloading a Book. In this case he was acquitted by the jury, but the article has some more details of how the court arguments went:

    The prosecution claimed Walker had deliberately retained his copy of [The Anarchist Cookbook] because he was “curious” about its contents. Walker denied this, saying he could not remember what he’d done with it after the game. Sellers, the prosecution lawyer, suggested Walker had endangered public safety by taking the book home and storing it in a drawer under his bed. […]

    Walker had two “reasonable excuses” that his lawyer presented to the court. The first was that the document had an academic purpose: It helped inform his university game, educating students about terrorism and counterterrorism. The second was that after the game had finished, he forgot that the document was in his possession.

    • Nornagest says:

      Is that the same Anarchist Cookbook that the FBI has treated as a joke for decades because you’re far more likely to blow up your own fingers than e.g. a mosque if you’re dumb enough to take its advice?

      • youzicha says:

        The very same!

        Broome said that she had worked for 25 years assessing explosives, sometimes forensically analyzing devices used in real terrorist attacks perpetrated in the U.K. and overseas. Bennathan, Walker’s lawyer, pressed her on whether she had ever encountered a terrorist case that involved the use of the “Anarchist Cookbook.” She could not provide any examples.

      • mupetblast says:

        There’s an interesting documentary on the author of the Anarchist Cookbook available on Netflix. William Powell – who lives in rural France with a somewhat obsequious wife – disavows his former radicalism but doesn’t feel responsible for murderers in the decades since its publishing who’ve cited the work or had it in their possession (James Holmes et al.). The filmmaker clearly does hold him responsible on a more than trivial level.

  48. James Miller says:

    At the time my son was born my wife and I had narrowed down his future name to either “Alex” or “Joey”. We went with Alex because he looked more like an Alex than a Joey.

  49. Dacyn says:

    Isn’t life expectancy at birth generally held to be a misleading statistic, because of infant mortality? The story likely isn’t about people dying at 19-23, but about (some) people dying much younger than that.

    • Watchman says:

      Yes, but normal infant mortality, when removed, is normally estimated in pre-modern socieities to give an average life expectancy of 30-35, so 19-23 suggests either calamitous levels of infant mortality, which is not matched by the longer life expectancy for women – to put it simply, high infant mortality increases the birth rate which reduces life expectancy for women – or an unusually dangerous society. Or that the estimations of female life expectancy are based on a model (I’ve not purused the paper) and I am building a historical case on a modern artifact…

      It should be noted this is not actually the whole population, but a landholding class by defintion, so this could be a pecularity of those in this particular situation with regards to land. If it was linked to military service for example, then the life expectancy could be understandable.

  50. Markus Ramikin says:

    That paperclip maximizer game is awesome. And no, I hadn’t played it or heard of it until reading about it here.

    Does anyone know how I can restart it? (I use Firefox, if it matters. Please make the answer understandable by a nonprogrammer)

    • Jack says:

      Delete the cookie for the website. In Firefox: Options -> Privacy & Security -> remove individual cookies -> search for the url and delete.

    • sandoratthezoo says:

      You can use the quantum thingie to drive your ops far enough negative and the game will give you an option to reset time and de facto restart.

    • Edward Scizorhands says:

      I seem to be stuck. I’m not making any clips any more (except what I click on to make them one at a time).

      Oh, I accidentally built solar farms ahead of schedule.

      • The Nybbler says:

        You need factories, solar production, harvester drones, and wire drones. And you need the Toth Sausage Conjecture (requires creativity) and Toth Tubule Enfolding.

        Edit: Oh, you’re talking about the third stage. If you’re stuck there you need more probes or more matter most likely. Bump up reproduction and hazard protection to get probes to survive and multiply, and give exploration at least a 1 to get matter. Don’t increase probe trust too much; it’s a trap.

        • Edward Scizorhands says:

          Am I really supposed to click on “Launch Probe” tens of thousands of times?

          I’ve already clicked on it 2.2k times.

          EDIT: Okay, things seem to be replicating well. But my “Total: ” is stuck at 5.71 thousand, despite that not being the net result. I think that’s a bug, because I see other people online with the same value.

          • Brad says:

            No. If you get the shielding and reproduction rates up to around 3 each to start you can get away with making just one.

            One non-obvious thing I noticed about that phase: when you change probe design it seems to change the existing stock of probes as well as new ones.

    • The Nybbler says:

      Open Javascript console. Type “reset()” or possibly “window.reset()”. I’m not a Firefox user so I’m not sure how to open the console. You need to delete localStorage for the site as well as cookies

  51. googolplexbyte says:

    Whales and Seals are closer related to Salmon than Sharks are so why not?

  52. J says:

    Exodus warns against bystander effect (Ex. 23:2): “Do not follow the crowd in doing wrong. When you give testimony in a lawsuit, do not pervert justice by siding with the crowd”

    And goes on to warn against bias either for or against people based on their socioeconomic status: “and do not show favoritism to a poor person in a lawsuit. … Do not deny justice to your poor people in their lawsuits.”

  53. Björn says:

    @Arabic scripts in surprising places: Even though the “arabic writing on viking cloth” theory might have been disproven, there is another historic garment that has very real arabic writing on it. It is the coronation mantle of the Holy Roman Emperor. Originally it was a mantle of King Roger II. of Sicily. Sicily of that time was a melting pot for Greek, Norman and Arabic culture, as it had belonged to the Byzantines and the Saracens before being conquered by Normans.

    There also is a Renaissance painting of the Virgin Mary(https://en.wikipedia.org/wiki/San_Giovenale_Triptych) that has the Islamic creed (shahada) written in her halo. The shahada often appears in muslim ornaments, and christian painters copied those ornaments, so by chance the shahada ended up in Mary’s halo.

    • John Schilling says:

      Sicily of that time was a melting pot for Greek, Norman and Arabic culture, as it had belonged to the Byzantines and the Saracens before being conquered by Normans.

      Sicily of which time, exactly? Because I’m going to guess that there was no “melting pot of Greek, Norman, and Arabic culture” in Sicily during the reign of any Emperor crowned under that mantle. The Emperor who inherited Sicily from the Normans, and presumably Roger’s mantle with it, was the same one who rounded up all of Sicily’s Muslims and sent them to a reservation on the mainland.

      Norman Sicily was a very successful experiment in high Medieval multiculturalism while it lasted, but it lasted barely a century (1091-1198).

      • Because I’m going to guess that there was no “melting pot of Greek, Norman, and Arabic culture” in Sicily during the reign of any Emperor crowned under that mantle.

        Norman Sicily was a very successful experiment in high Medieval multiculturalism while it lasted, but it lasted barely a century (1091-1198).

        Roger II was crowned in 1130, so that would be about when the mantle was made.

        Roger wasn’t a Holy Roman Emperor. Henry II was both Holy Roman Emperor and King of Sicily, however, from 1194.

        Frederic II, Stupor Mundi, was Holy Roman Emperor and King of Sicily, and his kingdom was sufficiently multicultural that Arabic was one of his languages.

        • John Schilling says:

          Henry VI, not Henry II, and he acquired Sicily when he was already Holy Roman Emperor, so he presumably wasn’t crowned emperor under the de Hauteville mantle. Frederick II would have been the first such.

          And Frederick would have learned Arabic from an early age as the daughter of Constance de Hauteville and as a native of twelfth-century Sicily. It isn’t clear that he had much use for the language as a thirteenth-century Holy Roman Emperor, except possibly for crusading purposes, and Sicily became noticeably less multicultural under his rule.

  54. J says:

    Odd that the articles say so little about Anthony Levandwoski’s motivations for starting a cult. My assumption is that he wants a place to stick his millions (and so much the better if it’s tax- and scrutiny-free) before his pants get sued off.

  55. Uncle Saturday says:

    Clicker games are basilisks for the hypofrontal; it’s cruel to link them.

    • Scott says:

      Agreed, I’ve found well made clicker games to be a more primally addictive and unrewarding time sink than anything else I’ve come across.

      • Edward Scizorhands says:

        I did get sucked in, rather noticeably. On the other hand, it was 100% free of ads and microtransactions.

      • Standing in the Shadows says:

        This is the very first clicker game I’ve ever played, which I think left me more unusually susceptible to addictive play. It chewed up more of a Thursday workday than I really could afford, as I had Important Documents due by EOWD.

        It took me just over 10.5 hours to convert the universe. I declined the Emperor’s offer, but it was a hard decision, which is Moar Paperclips?: all of this universe, or just most of this universe which then the drifters will convert back, and then all of another one? I picked all of this one, fuck those traitor drifters!

        Question: did buying the loadable strategies buy me anything? They made playing IPD slower and more expensive, but I don’t think I was getting any more yoni out of it.

        I never managed to use the Quantum Processors. Next time, I won’t even bother buying any.

        It was very odd to observe in myself the feeling of betrayal and anger as I watched the “lost to values drift” counter go up, and the feeling of just vindication every time I saw the drifter population smashed back to nearly zero after each major successful battle. It’s just counters in a javascript app, but it was more emotionally engaging than a huge expensive heavy on the art and “story” major video game.

        • Brad says:

          > did buying the loadable strategies buy me anything?

          It seems that these were tuned wrong and so ended up being counterproductive.

          > I never managed to use the Quantum Processors.

          Because the optimal strategy seemed to be to heavily favor memory over processors, I ended up using them quite a bit. When there all dark you could refill the operations bar quite quickly with them. Not too mention get a project 1-2 trust early.

    • Bugmaster says:

      FWIW, the original Cookie Clicker game was way better, IMO. It’s less addictive while also being more interesting.

  56. vV_Vv says:

    Eliezer Yudkowsky on the MIRI site: There’s No Fire Alarm For Artificial General Intelligence. There will be no particular event that creates common knowledge that it’s okay to say AGI seems near, so a conventional wisdom that it’s certainly a long way off will last long past the point when the evidence suggests otherwise.

    From the link: ‘I got up in Q&A and said, “Okay, you’ve all told us that progress won’t be all that fast. But let’s be more concrete and specific. I’d like to know what’s the least impressive accomplishment that you are very confident cannot be done in the next two years.”’

    Related…ish: Paperclip maximizer, the game. As if all of you haven’t already played this. If you don’t get the reference, this article explains.

    Ok, I’ll attempt a (humorous, but actually serious) reply to EY’s question: within the next two years AI will not be able to beat the paperclip maximizer game (or any other game of this type).

    To clarify: the game strategy, the game mechanics, or any game-specifc heuristic should not be programmed in the AI (otherwise it would be trivial to program a bot that beats the game). The AI should receive as input either raw pixels or some higher-level representation of the game UI (text boxes, buttons, etc.), and it should be designed to maximize the paperclips (indeed). It can have background knowledge of English, it can have access to Wikipedia, LessWrong, etc., but not to guides for the game. It must be able to beat the game within 10 playthoughs. I’m 99% confident that this technology will not be available two years from now.

    • Le Maistre Chat says:

      I was just thinking that the decision tree in Go is still a very limited abstract map of the territory “military strategy”. The Butlerian Jihad is not nearly as pressing as it will be if an AI starts winning war simulations.
      Furthermore, I wouldn’t call it an AGI until it beats the best Civilization players without cheating, then can have a conversation about how the game/map misrepresents the territory.

  57. Plucky says:

    Vikings possessing Arab artifacts should not be the least bit surprising. They were enterprising traders who for several centuries controlled both the Atlantic coastal trade to Umayyad Iberia and the riverine routes through Russia to the Black Sea and central Asia.

  58. esraymond says:

    Scott: I think Thrasymachus’s advice is utterly horrible.

    Extended rant on this at Against modesty, and for the Fischer set

    • Deiseach says:

      People misunderstand modesty/humility. It is not “I am smart and I know I’m smart but I’ll pretend I’m not as smart as I really am and I’ll pretend stupid things are not stupid”, it’s “Yes I’m smart but that does not mean the people who didn’t notice this before are idiots or that I’m necessarily smarter than they are, and being smart in one area does not mean I will be equally successful in other areas”.

  59. Nancy Lebovitz says:

    http://www.metafilter.com/170305/The-delay-doubled-every-time-a-wrong-PIN-was-entered

    Possibly fun– Wired editor secures his bitcoin(s), and discovers he’s secured them against himself. All seems lost until there turns out to a security hole in the security he bought…..

    If you don’t have enough computer aggravation in your life, you might like reading about someone else’s. The metafilter thread includes both sympathy and gloating.

  60. dvasya says:

    The interpretation of the fire alarm experiments seems somewhat controversial given that, as a matter of fact, there wasn’t a fire and the students who “failed” to react were, in reality, correct. The students who were alone in the room ended up much easier to fool! (Not much relevance to Yukdowsky’s points on the actual matter, just complaining about a reading experience partially ruined by attention constantly returning to this thought throughout the essay…)

    • anon1 says:

      Yes! You’d have to be really naive to participate in a psychology study, see something really unusual, and not suspect it was faked as part of the experiment.

  61. ellevt says:

    I get that container ships are not cruise ships, but they both burn bunker fuel which is very high in sulfur. Cars vs. ships was covered in the BBC’s More or Less (hosted by my favorite economist, Tim Harford).

    Local vs. global effects of bunker fuel pollution is covered in a few episodes of the excellent podcast series Containers

  62. Riothamus says:

    From the body camera article in NYT:

    “But the Rialto experiment featured just 54 officers, compared with over 2,000 in Washington. Officers in Washington captured five times as many hours of video. The larger sample size and the long-term way the cameras were assigned added to the reliability of the D.C. results.”

    This throws a flag for me. 40x the population of officers, over a longer time period, and only 5x the video? Is this a typo, or what?

    • Protagoras says:

      The reference to “long term way the cameras were assigned” leads me to assume the 5x hours of video was per officer, not total (so 200x the video total).

  63. hollyluja says:

    I wonder how much the name-looks thing is a function of class, money, or parental age. Older parents tend to name their kids more traditional names with standard spelling. Name trends in general filter down from rich to poor.

    I’ve noted that here in the US, as long as your parents were middle class, most people look pretty decent at 20. They got the orthodontia, the sunscreen, the acne treatment, the makeup and clothing to make the most of their genetic potential, looks wise.
    But when I visit my Aunt who teaches drama at a private wealthy high school in Nashville, the kids there are gorgeous. I have no idea what treatments they have access to that the rest of us don’t. Or possibly it’s genetics, with the rich marrying the beautiful or the beautiful becoming rich? It’s bonkers.

Leave a Reply