This is the…monthly? bimonthly? occasional?…classified thread. Post advertisements, personals, and any interesting success stories from the last thread.
Under the circumstances, you can also ask for financial support (eg link a GoFundMe) if you need to. Nobody has to donate to people who do this, but please at least be respectful.
Metaculus is a platform for generating crowd-sourced predictions about the future, especially science and technology. If you're interested in testing yourself and contributing to their project, check out their questions page
Jane Street is a quantitative trading firm with a focus on technology and collaborative problem solving. We're always hiring talented programmers, traders, and researchers and have internships and fulltime positions in New York, London, and Hong Kong. No background in finance required.
Substack is a blogging site that helps writers earn money and readers discover articles they'll like.
Beeminder's an evidence-based willpower augmention tool that collects quantifiable data about your life, then helps you organize it into commitment mechanisms so you can keep resolutions. They've also got a blog about what they're doing here
AISafety.com hosts a Skype reading group Wednesdays at 19:45 UTC, reading new and old articles on different aspects of AI Safety. We start with a presentation of a summary of the article, and then discuss in a friendly atmosphere.
Norwegian founders with an international team on a mission to offer the equivalent of a Norwegian social safety net globally available as a membership. Currently offering travel medical insurance for nomads, and global health insurance for remote teams.
The Effective Altruism newsletter provides monthly updates on the highest-impact ways to do good and help others.
Dr. Laura Baur is a psychiatrist with interests in literature review, reproductive psychiatry, and relational psychotherapy; see her website for more. Note that due to conflict of interest she doesn't treat people in the NYC rationalist social scene.
MealSquares is a "nutritionally complete" food that contains a balanced diet worth of nutrients in a few tasty easily measurable units. Think Soylent, except zero preparation, made with natural ingredients, and looks/tastes a lot like an ordinary scone.
80,000 Hours researches different problems and professions to help you figure out how to do as much good as possible. Their free career guide show you how to choose a career that's fulfilling and maximises your contribution to solving the world's most pressing problems.
Altruisto is a browser extension so that when you shop online, a portion of the money you pay goes to effective charities (no extra cost to you). Just install an extension and when you buy something, people in poverty will get medicines, bed nets, or financial aid.
Support Slate Star Codex on Patreon. I have a day job and SSC gets free hosting, so don't feel pressured to contribute. But extra cash helps pay for contest prizes, meetup expenses, and me spending extra time blogging instead of working.
Giving What We Can is a charitable movement promoting giving some of your money to the developing world or other worthy causes. If you're interested in this, consider taking their Pledge as a formal and public declaration of intent.
The COVID-19 Forecasting Project at the University of Oxford is making advanced pandemic simulations of 150+ countries available to the public, and also offer pro-bono forecasting services to decision-makers.
Seattle Anxiety Specialists are a therapy practice helping people overcome anxiety and related mental health issues (eg GAD, OCD, PTSD) through evidence based interventions and self-exploration. Check out their free anti-anxiety guide here
.
B4X is a free and open source developer tool that allows users to write apps for Android, iOS, and more.
I’m on a relatively formal discord that is focused on politics and philosophy, I’ve met some very intelligent people here, so feel free to join us. It’s not as formal as the SSC discord from my experience but in the channels specific to topics discussion can be of high quality.
https://discord.gg/s9xDCQk
Hi, I’m Jeff, and I work for NAQT, a company that produces quiz bowl questions. Quiz bowl is a game of knowledge that covers a broad range of both academic and pop culture subjects. For the last twenty years, we’ve been running championships to crown the best high school and college quiz bowl teams in the country. This spring, for obvious reasons, we haven’t been able to host championships in person. So we built Buzzword: a new online platform that allows anyone to play quiz bowl remotely. Buzzword is open to everyone (you don’t need to be enrolled in school or have an existing trivia team). If you enjoyed playing quiz bowl in high school or college, or if you never played but wish you had, check out our site: https://www.naqt.com/buzzword/
Thanks for posting this! I’ll check it out after work, my bar trivia team has been looking for something to do since the lockdown began
Thanks for checking out Buzzword! (Your team might also want to consult the calendar maintained at https://thejeopardyfan.com/2020/03/live-streamed-trivia-events-calendar.html, which lists a bunch of options for online pub quizzes.)
Hey, your database of questions would be useful for calibration apps; high-quality questions are one of the limiting reagents. You could create a calibration app to rule them all, or contribute to an already existing one.
Sparkwave.tech might be interested in doing that, and has the expertise.
What “learn programming online” sites and tools are good and free? Most that I can easily find try to push you into paid programs as soon as possible. (Something like codeacademy used to be would be ideal)
A note on codecademy, overall it’s not that great.
It’s well designed to make you feel like you’re progressing, but it’s hard to apply the lessons learnt in their environment to real projects. It can be good for languages that you may only need a basic understanding of (like SQL for many devs), but generally you need to supplement it with some more in depth study. Duolingo maybe a good comparison.
If you are completely new to programming, I would strongly recommend reading the SICP, and doing the exercises. It won’t teach you anything about writing practical code or getting paid to do so. It won’t teach you to use any modern programming language, either. So, from a pure ROI point of view, it’s useless. However, what the book will do is give you the basic grounding in CS theory, thus empowering you to learn any programming language or system about 10x faster than you otherwise could.
After a lot of success with using anonymous Twitter alts for self-therapy, I want to bring a similar concept to blogging and site-making.
* It should be possible to make a new blog/site/project page in one click.
* It should be possible to keep all your pages anonymous and distinct from each other.
* Publishing should be as easy as writing in a Google Doc. No Markdown, no git repos, no “save” button.
* There should be no ads (I detest ads).
So I’m building a service for that β Brick.do. It already works and you can use it free of charge. (I might start charging for subdomains later, but it’ll be possible to use without a subdomain.)
I’ve been using it myself and I can report that it already makes writing significantly easier than other publishing platforms, by removing pretty much all friction from the process. I wrote several posts in a week, while previously I was doing one in several months.
A friend of mine who lost his job just before lockdown has spent the time finishing developing a card game that he’d been working on for years. I’ve played the test and it’s really good. He’s raising money on Kickstarter to do an initial batch, and he’s just over halfway to the goal: https://www.kickstarter.com/projects/soyeahimjames/days-to-ragnarok
I built a service for managing loans between friends after a really shitty experience trying to get my student loans refinanced. The goal is to remove the social anxiety of *servicing* friends and family loans that gives them a bad reputation.
https://getsima.com/
I suppose I could link again to my DeviantArt gallery. I mostly work on the branch of speculative biology, the creation of imaginary organisms and worlds that follow, as close as possible, the known laws of biology — to quote my main page, “The results of evolution at work in other worlds and realms of being, in the far future of our own planet, in timelines that diverged from the one we know, in unexplored angles of the deep past, or by hand of unscrupolous Creators”.
I’ve also made a few infographics, mostly pertaining to real-world biology. I’m especially proud of my summary of human evolution.
I’ve drawn fictional creatures and maps (for example) for works and projects by other authors, including a couple among our own fellow commenters. I have also drawn real-life anatomy and diagrams for biology textbooks, one to be published by the American Museum of Natural History. I’d be willing to take more commissions of this kind, if anyone were to need one.
Took a look at your summary of human evolution, and it’s really impressive. Thanks for sharing. I remember reading a book as a kid that contained speculations on what future animals would look like. One possibility they considered was that descendants of squids or octopi would evolve to become land dwelling creatures, walking around with tentacles adapted to support their weight. I think the book also had predicted maps of the world and its climate based on plate tectonics. I don’t know about the climate and the plate tectonics, but I doubt their speculative animals are likely to evolve. It was a very interesting book, though, and quite similar to the stuff you are doing. I wish I could remember the title.
Thanks!
Sounds like you’re talking about The Future is Wild. You can still see most of its creations here, and you can probably find the TV series somewhere. I don’t see anything obviously wrong with the shape of continents (the disappearance of the Mediterranean and the separation of East Africa are generally expected by geologists, as is the formation of a new supercontinent 200-250 million years in the future), but you’re correct that their specific predictions about life are less than perfectly plausible. The authors have tried to make their animals as surprising as possible by wiping the slate clean with implausibly wide mass extinctions and evolving replacements that often don’t follow all evolutionary constraints. Still a very imaginative and entertaining work.
In you’re interested in the genre, I recommend After Man by Dougal Dixon, whose books were discussed a bit last month.
Since shameless advertising is permitted here, I am currently looking for a programmer that like machine learning.
I.e I need someone to work on an ML project, but mostly on the non ML bits, however, some knowledge of ML will still be required in order to not fuck things up and to be able to properly review code, scrutinize features and test things.
The only requirements I have is some basic understanding on ML and being a good~ish programmer that knows python, also bonus points most thing you’d be working on will be open source and the company has a ~4M seed, so we can keep the thing going for 2-3 years even in a worst-case scenario.
The job is remote, which means that if you live in 99.9% of the world we’ll probably pay you better than average, but if you happen to live in the 0.1% of cities which are very inflated because all intelligence workers moved there in order to have their laptops sit nearby each other, then I probably can’t afford it.
If anyone’s interested ping me at george@cerebralab.com
Already found enough people, please ignore this, thanks to everyone that emailed me π
Probably missed the window where people are still reading the comments, but here goes:
My friend and I made an online word game, alphabattle.xyz. You can play on your own (against a dictionary) or against others.
Basically, after guessing a few words, you get a puzzle like below and need to find the word that matches.
DREAM β0 β0
CODON β0 β2
TWINK β0 β1
CLOUD β1 β0
BROTH β2 β1
THONG β2 β1
It’s free and I’m advertising it mainly because I need opponents π Potential side effects include turning into a human RegExp machine and being unable to read without noting down unusual five-letter words.
People need on average 12 words to solve, so don’t give up too quickly. Also, the English dictionary currently contains many ultra-rare words. Sorry if you get a very tough nut. Or challenge me, English is my not my first language..
If you’re ready to start programming, check out my book C++ for Lazy Programmers (https://smile.amazon.com/Lazy-Programmers-Quick-Easy-Beginners/dp/1484251865/ref=sr_1_1) and my YouTube channel Programming the Lazy Way (https://www.youtube.com/channel/UC2pJX9Oq5qDgLgTeMnZbDsg).
The book may have a new edition by summer’s end, updated to the latest (C++20) standard.
I’m just getting started writing fiction, and am interested in joining any writing groups that might be in the SSC-verse. My own novel is near-future cognitive science-themed scifi (BCIs, AGI, psychology, etc). Message me at tyleralterman@gmail.com if you know of anything good!
My book is being published on Tuesday: Agile Conversation: Transform Your Conversations, Transform Your Culture
The heart of the book is how to analyze your conversations to change the way you communicate, with each chapter giving you a different tool to measure yourself against.
Jtf
Anyone in the market for a nice, affordable, ocean-worthy sailboat?
See here
After working on it for an extended time, I have now started posting a series of articles that seeks to give a naturalistic explanation of what’s going on with Buddhist insight meditation, enlightenment, and weird claims like “in order to get enlightened you have to stop striving for enlightenment”.
I recently posted the first substantial part in the series, discussing the concept of “no-self” and how the experience of a self is built up and how meditation affects that: A non-mystical explanation of “no-self”. I have several more posts written up and scheduled to be posted. Several people have already told me that they liked this article a lot. (There’s also an introduction where I more generally cover my approach and background assumptions etc.)
This is part of a series where I have been putting together my own synthesis of how the mind works; my review of Unlocking the Emotional Brain, which Scott referenced in his Mental Mountains post, is a part of the same series.
Hello! This is a repeat from the last classified thread, if you’ve already read feel free to skip; it’s basically just the same information. Posting anyway because there always seem to be new people reading. Now on to the ad…
I am a freelance editor specializing primarily in fantasy and science fiction, though happy to work with most types of fiction (the more of a genre I’ve read, the better I am; no sex scenes please, I’d be utterly useless; that said, I have no problems at all with stuff that doesn’t quite fit in any genre, or is generally strange – I enjoyed Unsong) and occasionally interesting non-fiction. My previous work includes fantasy, light romance, the variety of not-quite-fantasy where the geography and history are invented but no magic is present, superheroes, urban fantasy (you may be noticing a pattern here), military sci-fi, mysteries (usually fantasy mysteries), and “interesting nonfiction” (mostly my father’s books, that’s how I got started). Not all of this is published, but for some representative samples: Harald (see note about how I got started), Cantata (a really early one which I firmly recommend on its own merits), Curveball (web serial superhero fiction, some of the editing I have done is in visible comments, though not all), and new this year, <a href="https://www.amazon.com/gp/product/B07PGJBQHT/?ie=UTF8&redirect=true&ref_=series_rw_dp_sw"Pirates and Powers (I belive I picked up this one from previous SSC ads – fun anime-inspired serial).
My own website is here. Prices are currently at $2-$8/500-word page depending on how much work is needed (for fiction; non-fiction is higher), with a five-free-pages offer for new clients so you know which of those prices applies and what you’ll be getting for it before you actually have to pay me anything. Contact information and additional details are all on the website. And thank you again to everyone who contacted me after the previous ads!
Hello, everyone!
My wife has finished a rough draft of a four-book YA fantasy series, and she’s in the process of editing her series now. She intends to self-publish for intellectual property reasons–she has poured a lot of heart and soul into this series, and she doesn’t like the idea of a big publishing company getting any say in its creative process or presentation. With all that said, do you all have any advice on the publication process? Do you have cover illustrators you recommend? Wholesale providers? Promotional strategies or resources?
– You can find freelance book editors!
– I started a bulleted list but I don’t actually have any other tips. You can get a professional editor without going through a publisher, though.
Hi there! My sister is a successful self-published YA fantasy author, so while I don’t have first-hand experience with self-publishing, I have had her ramble at me about it for hours. I’ll just cover what my sister did, which I assume is fairly good, because it’s gotten her to a place where she was able to WIN the 2019 SPFBO (self-published fantasy blog-off), which is still so crazy to me.
Publishing process:
– She publishes through Amazon, both Kindle and print-on-demand paperback. While the Kindle version costs much less upfront, authors can get paid based on how many pages a person reads of their book. Going through Amazon seems like a clear winner to me β the books are high-quality, if not quite what you’d get through a traditional publisher, and they handle distribution for you.
– She did her own cover art, so I don’t have any insight there, sorry. I can say that you will want to iterate on design (both inside and outside) a bunch before releasing it β Amazon can send you preview copies β because things will look pretty different in-person than they do on a screen.
– She didn’t use a professional editor (because even that felt like relinquishing too much creative control); instead she gathered a group of friends and acquaintances who were interested in reading her books and sent them a few chapters a week. That way they get to feel cool for reading the book before it was published while also providing valuable feedback. Ideally some of these readers would be in the target audience (e.g. teens) while others would have a bit more editing experience (e.g. my mom is a career English teacher, so that worked well for us).
Promotion:
– Make a mailing list! People like ~monthly newsletters, especially if they include sneak previews or book giveaways. My sister picks someone at random off her mailing list each month and sends them a free signed copy of one of her books.
– I know she spends a lot of time optimizing her Google Adwords. I think there’s something similar on Amazon where you can pay to have your book recommended from the pages of more popular books in the genre. I don’t know much here. Probably don’t give the book a title that’s too generic.
– Use social media, especially Twitter. This recommendation goes against everything I believe in because I hate Twitter with a passion, but it seems to help. Generally be available and communicative with your readers, even if you hate it. Making a website/blog is good too.
– Make friends with other authors online. This was HUGE for my sister. In her case, I think(?) what happened was that she contacted some authors and online book-reviewers to give honest reviews in exchange for ARCs, and one of them really liked it and started telling everyone she was the best thing since sliced bread. He acted as a sort of mentor to her in the world of self-publishing, which seems really great if you can get it. Notably, she was first ‘discovered’ in this way by other Asians in the book community, who were excited about her East Asia-inspired fantasy world and I think also sort of thought of her as one of their own and wanted to see her succeed. To generalize this, if there’s anything niche about your wife’s book, she might want to try finding and befriending people in that same niche.
– Perhaps a long shot, but once published, enter the SPFBO. They accept 300 entrants each year. Being a finalist is of course amazing for publicity, but even just entering will get you some exposure.
That’s all I can think of off the top of my head. If you have specific other questions you can email me at [thisusername] at uchicago dot edu, and I’ll pass them along to my sister. Time was I would have put you in touch directly but I bet she’s all busy now that she’s famous and successful (not actually sure that’s true I’m just very proud and still so excited; the contest only just ended last week!). Best of luck to your wife!!
I can’t offer anything constructive, really. I do know an illustrator, but AFAIK she has a steady job now, so I’m not sure if she would commit to anything on a long-term basis. That said, if you’re looking for someone to give your book a brutally honest review, I’m your guy π
If you’ve got a Mac, Vellum is a program for book design (typesetting and all that). You could do that in word, but this program makes everything so much faster. It is a bit expensive, but at least it sells a license instead of a subscription model.
Grammarly and Prowritingaid are programs for editing, which can be used to fix some grammar mistakes. Won’t fix everything, but helps send editors a cleaner version (the more time they spend on it, the more they charge).
My gaming headset just broke, so once again, I’m looking for recommendations. All of my requirements are fairly ordinary… except for one:
* USB connector (which implies that the headset is its own sound card).
* Good mic.
* It has to be comfortable to wear for prolonged periods of time, for someone who has a big fat head.
* Good sound isolation.
* Sturdy construction (my current headset failed this test).
* It needs to present itself as a 7.1 audio device in Windows, and it needs to be capable of actually delivering decent surround sound. I need to be able to tell if the enemies are shooting me from 11 o-clock, or from 7 o-clock.
* Just to reiterate, I’m not looking for a regular headset with some kind of a “virtual surround sound boost xtreme” mode, but a device that Windows (and thus, the games I play on Windows) would believe is a 7.1 speaker system.
So, any recommendations ?
I’ve been using a Steelseries Arctis 7 for a while now, after I had two older headsets fail the “sturdy construction test” same as you. My wife also bought the same pair after one of our cats chewed through the cord on her old pair, and she seems happy with them as well. You didn’t meantion wireless but that’s a nice added benefit that I can’t imagine going back to not having once I’ve experienced it, and I’m able to get literally a full day and a half to two days of gaming in before it starts giving me a low battery warning (And even then it’s still perfectly usable as a wired headset while charging (USB-A to micro USB, cable included but I opted to use one with a detachable magnetic head I had lying around). Wireless range is honestly incredable, I’m able to wander basically anywhere in my home, including out to the front to take out the garbage and the like, and still hear just fine as long as there’s less than 3 sets of walls between me and my computer.
It does connect via USB (The wireless receiver does, I mean) and the mic is good enough for gaming but probably falls short if you’re going to be podcasting or doing serious streaming or anything like that. The sound isolation is good enough that my wife sometimes gets a little frustrated that I didn’t hear something she said. The product page does list DTS 7.1 surround sound, and I can confirm in my system tray right now that Windows does see and recognize this, and gives me the option of turning it on or using Windows Sonic as well.
Other neat feature: It actually shows up as two separate audio devices to windows, the idea being you can set your discord/skype/whatever to one and your game to another, and adjust the volume mix between the two the fly with a small wheel behind the right earcup (The wheel behind the left earcup controls volume overall). You can ignore this entirely if it’s not to your liking by just setting all your programs to use the same audio output, same as normal.
If I one complaint, it’s that the retractable mic (which is better quality-wise than anything I’ve used before) doesn’t quite extend out as far as I’d like, speaking as someone who also has a big fat head. It’s still very usable, but it only reaches to maybe the corner of my mouth where older headsets I’m used to extend maybe 80 degrees around my head instead. It’s a small nitpick and I’m perfectly happy with it overall.
Thanks, I’ll check it out. I normally stay away from wireless headsets, because in my experience they inevitably introduce lag and all kinds of glitches — but maybe Steelseries finally hit on a winning formula…
Yeah that was my experience too, but it seems the state of wireless peripherals has been advancing steadily while I hadn’t been looking. Wireless mice are already as good as their wired counterparts. Wireless headsets are imperceptibly behind, and top wireless keyboards are in “almost good enough” territory.
I saw other people posting their music, and I think it’s cool, I like to see fellow musicians on SSC. So here’s my stuff: https://kruasan.bandcamp.com
Most of the time I make something electronic with a lot of funky melodies, but sometimes I make ambient, rock, orchestral, drum and bass, whatever. I’m bored, so if you want to hire me to make music for your thing or you just want to collaborate – let me know.
Is your organization looking to hire a mid- to senior-level UX (user experience) researcher?
I’m one of those! I have 7 years of experience, mainly with employee-facing and B2B products, and in the past few years I’ve specialized in qualitative research supporting change management initiatives at large companies. Ideally I’d like to keep doing that but I’m adaptable and can learn quickly if the chance to do other kinds of UX research arises.
I’m a strong writer with an eclectic background. I was trained by a trio of Human Factors PhDs, who instilled in me a love of process improvement (particularly with Agile methodologies, which they were fairly purist about) and respect for scientific rigor (which they were definitely purist about). I hold certifications in qualitative research, human-computer interaction, and DACUM workshop facilitation. Through my work I’ve gained familiarity with the domains of electromechanical systems design, healthcare IT, electric utilities, and banking. I’m very interested in performing UX research for physical products and continuing to provide data for change management teams.
My resume, portfolio, and contact info are on my website along with my blog, where you can see some of the UX-related things I think about on my own time. If you’d like to reach out directly, you can email me at nadavzohar at hotmail dot com, or on my Linkedin page.
Thanks for reading, and thank you, Scott, for hosting this thread.
I’m a Data Scientist most recently working at a leading machine learning company with 7 years experience. Due to layoffs, currently looking around now for a new position (preferably in Boston).
Most recent specialization is explainable modeling, but have also worked in pharma, IT, and energy. Most dev experience is pythonic.
If you know of any good DS positions, try thomas.stearns26@gmail.com
I’ve flogged Pyrebound on here before (just click on my username to get to it), figured I’d give it another go-round. PB is an ongoing hard fantasy web serial, slowly winding its way towards a conclusion after a year and a half. Set in a world loosely inspired by ancient Mesopotamia where a cosmic disaster strikes once every four days, and human society has adapted in ghastly ways. It’s reasonably popular on r/rational. Thanks for reading this!
I’m trying to finish a book I’ve been working on for years and have lost steam on in the last year.
I am looking to crowd edit the overall format and vibe as I go.
The book is about the “Just World Fallacy” which was “discovered” by social psychologists 50+ years ago, but has received relatively little interest compared to a lot of cognitive biases. (One reason is that it could be more fundamental cognitive biases manifesting as a belief in justice.)
I have the rough Table of Contents here: https://justworldfallacy.com/table-of-contents/ and right now I’m looking for feedback of the outline plus pushback on any of the parts I’ve published so far.
If you’re interested but don’t want to comment publicly, you can email me at riley.haas@gmail.com
I know it’s still very much up in the air whether choirs will be able to perform (even livestream) this December, but: if your choir is looking to put together a December program, and you are looking for newly composed, seasonally appropriate music, I recently finished a piece you might like.
The piece is a setting of the Advent antiphon “O Clavis David.” It’s interesting and offbeat enough in style (5/4 ostinati, modal harmonies) to sound fresh, but simple and accessible enough to work for amateur and college choirs or even a somewhat ambitious high school choir. I figure that the lines “Come and free the captive from the prison house, who sits in darkness and the shadow of death” might resonate emotionally with a lot of people right now.
The piece is available for purchase at https://nicholasweininger.com/scores, and I am happy to send a free perusal score to anyone who would like; email me via the form at https://nicholasweininger.com/contact if you’d like one.
Not directly relevant to your post but an irl friend of mine has started a platform for online choir events called Click and Sing who might be of interest https://clickandsing.co.uk/
My company just finished the most recent batch of YC and is looking to bring on a strong biologist or chemical biologist to found/build out our biology team. We are making the next generation of PROTAC-like molecules to go after previously undruggable proteins. Shoot me an email if you’re interested or would like to learn more βΒ armand@weatherwax.bio π
I’m working on a dashboard that shows Covid-19 stats for the county where I live (in the US). There are some existing dashboards out there, but they don’t present the data in a way that makes the most amount of sense to me. In particular, the existing dashboards don’t show the trend in newly-diagnosed cases, which seems like a critical piece of information. (I used Shiny because my original plan was to make it interactive, but the interactivity seemed like it didn’t really add any value, so I dumped it).
I’m interested in any feedback about ways to improve it. It’s a work in progress.
My universal gripe with popular covid19 data reporting is using absolute numbers without context – cases should be displayed per 1000 people in the county, a comparison with surrounding counties and hotspots like NYC can be useful for perspective, the data about hospitalizations and patients on ventilators will make more sense when displayed as percentage of local hospital and ICU capacity.
I started a blog in 2019, where I do data driven analyses mostly of scientific questions and occasionally just ramble. A big motivation for many of the analyses was to check on my own, whether the bad men on the internet are correct with all the nasty stuff that is anathema to my very liberal upbringing, so it is quite cw-ish.
https://halfassed.science.blog
The name is my way of hedging that I am low on conscientiousness and time, but I try to make up for these shortcomings with interesting ideas.
I did a series called chess psychometrics where I try to answer psychometric questions on the basis of chess databases.
I did a prove of concept for the idea that one might detect the increased use of HGH in high-level sports by checking whether face embeddings of Olympic contestants these days are shifted into the direction in which people with pituitary tumors are shifted compared to earlier Olympics.
I present one idea how to create nootropics that really work, ie. increase IQ not just focus or endurance.
I have one theory of IQ that explains the Flynn effect and several other observations in the field.
Etc.
Currently 50+ posts. Did weekly updates in 2019, now more like monthly.
Link to the Flynn effect theory?
My company, Wave, is hiring a “Security Communications Lead”. We’re trying to beef up our internal process around operational security, and as part of that, we need to deliver process changes to the whole 200+ person company.
The usual sourcing methods haven’t been working — but it seems like the sort of skillset that might be represented here on SSC. We’re looking for a great communicator with a technical mindset — not just detail-oriented, but good at understanding technical stuff. You don’t need any technical experience in particular, but you need to be able to interview technical people and understand tradeoffs, rationally think through scenarios, make good tail-risk estimates, then improve the process. Think through about what processes will *actually work* to get a very diverse team of people to not leak important data, etc.
Wave’s been mentioned on SSC once (“Silicon Valley: A Reality Check”). We’re a for-profit org staffed by some effective altruists, working on mobile money and financial inclusion in Africa.
Job description and application form here, https://boards.greenhouse.io/wavemm/jobs/4681943002, but you’re welcome to also shoot me an email (lincoln@wave.com) and I’ll make sure it gets seen. This job is fully remote except (after COVID subsides) travel every few months to Africa. It’s best if you can overlap timezone-wise with West Africa (UTC.)
Hi! I have a Bandcamp page here: Ce N’est Pas De La Musique There’s a lot here: ~20 albums with almost 20 tracks per album. Very little vocals. The albums were modeled after traditional vinyl albums; two sides, approximately 20 minutes per side, for a total of about 40 minutes of music, more or less. Each album is roughly designed to have a flow and dynamics so as to create interest and are not necessarily a collection of individual songs.
Iβm looking for listeners, and hopefully some constructive feedback. And while I donβt like talking about my music, I donβt mind if you do.
I would appreciate any comments, ideas, suggestions, etc.
Thanks!
A frequent reader I’ve posted a bit on recent threads (and a few times before under a pseudonym less closely connected to my real name than this one). I’ve just started a blog, in the first instance about Augustine and his times, but potentially to extend to the wider late Roman world. There’s obviously not much there yet, and I’m wondering if anyone has anything they’d be really interested in hearing more about (or, for that matter, discussing in open threads).
I am interested in politics (especially government structure) and economics in Late Antiquity! I would also like to see the primary texts which reference the people listed in the Roman Canon assembled in one place. I am also interested in the use of economics metaphors in theology (Jesus, Paul, and Augustine deploy them).
I have some posts on the lineup about Augustine’s expectations on money-making, if not as yet about ancient economics as such (a highly technical subject about which I know mostly generalities).
The Roman Canon is an interesting angle. It would quickly wrap us up in questions (equally technical; I just happen to know more about them) about how to approach hagiography–and that might be worthwhile in itself. Hmm. Expect a post on this in the next several days, though probably not a complete listing of hagiographical texts, which would be exhausting to produce (how do you navigate all the medieval stuff?)
Sponge here. I soak up technical subjects with glee and live attached to the depths of my curiosity. Unfortunately, waters are shallow these days and I am in need of sustenance.
Not sure I can push the metaphor much further. If you need a developer or someone to organize, fix, and/or manage your computer infrastructer, I am your guy!
My background is in pure math, sysops and software development. Here is my (rough) CV: https://wilsonb.com/cv.pdf. I have ~5 years experience in industry and ~20 tinkering on Linux. Current projects:
– Add support for automatically deploying Google Cloud Platorm instances running a Guix System,
– Array-oriented text editor. Think line-oriented text editor like “ed” but without making newline characters special.
– Learning formal proof systems by implementing Hurwitz’s theorem in Metamath.
Languages I like: J, Haskell, Scheme, POSIX shell, x86 assembly
Languages I know: Python, C, Java, Objective-C
This isn’t exactly what you are asking about, but should that “Hierloom Bookshop” be “Heirloom Bookshop”?
I write about life (health, wealth, love, meaning) on my website on https://kaue.me
I design various things from business cards to wedding invitations(and wedding postponement/cancelation announcements) to home decor to electronics cases face masks. (most of them are made to be personalized with names, monograms, etc..) See my Zazzle store:
Retro Vintage Store
I do the SSC podcast (do you support it on Patreon, well do you?) and I suddenly find myself with a lot of free time at the moment so I’m trying to spend more of my time on sound and music projects.
I’m particularly interested in doing more video game sound and music design and implementation, sound design and dialogue editing for apps, animation or data visualisation (film too but not preferred) and podcasts. I’m also really interested in any kind of narrative audio fiction. Also composing music in general, as well as producing, mixing and editing it.
Thought it would be a good idea to put out feelers here since in my experience SSC people are generally my kind of people and are usually working on projects I can get behind.
I can send over examples of my work if you contact me on Discord at SolenoidEntity#7425 or you can email slatestarpodcast [gmail]
Please help recommend me a new laptop.
Right now I have a Surface Book 2, which is great but a couple years old now and starting to accrete minor problems. I almost never use the 2-in-1 capability. I feel like technology has advanced enough that I should be able to do better. I was waiting for Surface Book 3, but it seems underwhelming.
My main uses are:
1. Playing Civilization 4. I know this is a really old game, but turns still take forever on really large maps. I feel like now, 15 years later, with almost unlimited budget, there ought to be some laptop out there which can play even the most absurdly colossal Civ4 map lightning-fast, and I would like to find it. I think Civ4 is a single-core-only game, so more cores probably don’t help here.
2. Videoconferencing for work. It doesn’t have to have amazing video performance, just have a webcam and be reliable.
3. Blogging. This isn’t performance-intensive, but a good ergonomic keyboard would be nice. I use an external mouse, so I don’t care about the trackpad.
I’d prefer something about 15 inches. Size/battery/portability are not really issues (except in the very weak sense that I don’t want a desktop). Price is not an issue. I refuse to get a Mac.
I’m considering Surface Book 3, Dell XPS, Razer Blade, ???. Any suggestions welcome.
I bought this laptop (https://support.hp.com/ie-en/document/c05532253) about three years ago and it’s still working pretty well for me. The integrated webcam is pretty hi-res. It’s able to run quite processing-intensive games like Alien Isolation or X-COM quite well with a fairly consistent framerate, and running older games has never presented an issue. I generally use an external USB keyboard, so I can’t really judge how good the integrated keyboard is, but on the rare occasions I have used it, it’s never presented any noticeable issues. Hope that helps.
The XPS laptops are very nice, very few laptops on the market match them in build quality. I’d guess that Civ4 is probably either CPU- or memory-bound, so if money’s truly no object you could get one of the top-end XPS 15 models. The two most expensive ones have the i9-9980HK which is probably the most powerful CPU you’ll find in a laptop that isn’t practically a portable desktop*, as well as ridiculously excessive amounts of RAM. They also come with the GTX 1650 for discrete graphics, which is probably more than good enough.
I think Lenovo are probably worth looking at as well in this category, but I don’t know what to suggest off the top of my head.
Otherwise you could look at bigger, chunkier laptops, basically either workstation laptops or edgy gamer laptops. Chunkier laptops get to have better thermals, which can support marginally better CPUs and allows the CPU to perform closer to its theoretical peak, and also supports substantially more capable discrete graphics chips that are more power-hungry. I doubt Civ4 cares enough about the better graphics performance, and you’d be getting diminishing returns on the CPU performance in exchange for the inconvenience of having a chunky laptop, so I’d probably just stick with something like the XPS 15.
*At least short of the Ryzen 9 4900HS, which seems to only be available in the Zephyrus G14 right now. It seems like a nice laptop, but unfortunately they left out the webcam.
I’m not sure that you’re right about Civ 4. I am not a computer guy, but if the architecture is built for the computers of 15 years ago (and if it’s single-thread only, then it was) the improvements in modern computers may just be in ways that don’t help. For instance, clock speed haven’t risen that much, and if that’s the bottleneck, then there’s nothing you can really do.
I might install my copy (which I haven’t played in a long time, and would be tempted by if Aurora hadn’t just come out) and see how my machine handles it.
(Also, what mods do you play with? I mostly did Total Realism/Realism Invictus).
I know Scott used to play Fall from Heaven II, at least.
Fall from Heaven and Pie’s Ancient Europe. Everyone already knows the first, but I also can’t recommend the second highly enough if you’re interested in the classical world or just like good Civ mods.
May have reported this comment by mistake, sorry about that.
Which report button did you use?
The red βReportβ button.
It’ll probably be ok.
Careful, you’ll get Scott banned.
Can’t help you that Scott, sorry, but since it’s on-topic for the thread let me point out that you might not have seen this last FfH story I wrote.
You’re Nikas-Knight??? Thank you so much for your work. You’ve helped take many hours of my life from me and I have no complaints.
EDIT: Wait a minute, I know Scott is Yvain on SSC. He’s not Yvain from FfH, is he?
Hey thanks. And no, though I do believe the LW screenname came from there.
I recently bought the Lenovo C740 Yoga (14in screen, 10th gen I7, 16gb RAM, 1 TB SSD). I use it for writing, reading, videoconferencing and programming (R data analysis mostly). So far, I have liked it quite a bit. It is fast, able to handle all of my needs easily, it has a very responsive keyboard for a laptop, and is very light and portable. The case is also pretty durable, which was a big factor for me (my previous laptop was an ASUS Zenbook, which apparently is notorious for its faulty hinges…) Price was also a factor as this laptop was a little over $1000, in comparison to >$1500 for a similar Surface or Dell XPS.
This is what I’m using: https://www.asus.com/Laptops/ASUS-ZenBook-15-UX533FD/
I bought it a year ago, for $1400. 15.6″ screen, 1 TB solid-state drive, so reliable and lots of storage. Dedicated graphics card and 16GB of RAM, so it will play games like Civ4 very well. *very* light-weight and slim for it’s size- often marketed as “the smallest 15″ laptop”.
Moreover, I’m happier with ASUS than I am with most big manufacturers- better customer service, reliability, etc. I use some of their hardware in my desktop as well.
EDIT – I should add I work a fulltime job 100% remote, using this laptop. Webcam is quite good for a built-in.
Just chiming in to say that I also really love Civ 4… and I don’t think you are going to get it to run much faster. The most obvious way to make a program run faster is by raising CPU speed, and this will improve program run speed pretty much independent of how the program is written. During the 90’s CPU speed rose very quickly, so that a 2000’s computer could run an old game way better than a 90’s one. But CPU speed hasn’t been going up very much over the past decade or so – the gains to computing power are coming from more esoteric places that older software can’t really take advantage of. A lot of it is simply that we are getting institutionally better at programming. And a lot of it is coming from parallelism, but as you noted, Civ 4 is single-core and wasn’t written in the time of using a graphics card for anything other than graphics, so there isn’t much you can do on the hardware end to get it to run faster.
My intuition is that the biggest thing that slows the game down on large maps is computing the unit AI. It is possible that somebody is developing a more modern AI system that could massively improve the game’s process speed. Any other fans out there have an opinion on this?
Also, what version / mods do you play? I mostly play Rise of Mankind, A New Dawn, with a couple of custom tweaks to the balance.
That’s true iff by institutionally you mean having learned to abandon attempts of clever coding for speed and developed optimizing compilers that do it better than us.
Ask people in gaming forums about civ4, it is possible that for your definition of ‘absurdly colossal’ there is just nothing on the market – laptop or desktop that will play it lighting-fast simply because of how it was designed. As other people here mentioned, the sequential (single-core) computation throughput for CPUs for last 15 years has been rising only modestly and even regressing for energy-conscious devices such as as laptops.
I’m currently on a year-old Dell XPS (13inch version, running Linux). I definitely recommend it. If you want battery life >2 hours, buy the low-resolution version. More details:
– I bought the low-resolution version (still a respectable 1920×1080 on a 13inch screen). My battery life is roughly ~12 hours of normal work, 2-3 hours of video/gaming/streaming. My friend with the high-res version generally gets only 1-2 hours of battery regardless. The low-res version is also noticeably lighter than the high-res version. I generally recommend buying the low-res version, but it sounds like the added weight and lowered battery wouldn’t be an issue for you.
– It’s also quite slim and surprisingly light, given its power.
– The webcam quality is as good as I’ve ever seen in a laptop.
– I can run “light” to “medium” video games without a problem; I would expect it to handle Civ 4 well, but checking a gaming forum for confirmation is a good idea.
– The speakers sound sort of tinny when they play music, but weirdly hearing voice through them does *not* sound tinny.
– If you type while on voice chat using the built-in microphone, people will definitely be able to hear clicky-clacky typing noises.
– I like the feel of the keyboard. The most similar keyboard I can think of is a mac keyboard, but this one has slightly stiffer keys with a different texture.
– You’ll be dragged along to the USB-C revolution whether you like it or not, definitely make sure you buy the relevant “””dongles”””. I always carry a converter to HDMI, VGA, and USB-A.
I know you said “Size/battery/portability are not really issues (except in the very weak sense that I donβt want a desktop)”, but the big hot new thing in desktops is really small form factors, some computer cases are so small they fit in a backpack and still have full-power high-end modern desktop parts. Of course, monitors are still an issue and portable monitors are generally pretty bad. Still something to look into maybe.
The new Dell XPSes are good, The Razer Blades are fine. Ultimately they’re close enough and it’s going to come down to intangibles like which keyboard feels better to you in particular. If we were living in different times, I would tell you to go out and try them out in person before buying them, but alas that’s impossible. Maybe you could order them all and return the ones you don’t like? Or get them one-by-one and return them until you get something acceptable? That’s normally an abnormal thing to do, but you did say “price is not an issue”…
You’re very correct to refuse to get a Macbook.
PS as a fellow Civ 4 player, I’d like to take this opportunity to recommend the Civ 4 mod “Fall From Heaven 2” to anyone who has ever played the game before. It’s supremely excellent. Predicting with >95% certainty that Scott already knows about it, but someone else reading this might not.
+1 for reconsidering a desktop.
A desktop and a Chromebook is a good combination, which you can get for less than a Macbook Pro. This will also allow you to get unrivaled cooling and single-threaded performance, which – as has been pointed out – is all that matters for Civ4 performance.
In my experience, an $20 external webcam is far superior to any standard laptop camera for video-conferencing. If that would work for your needs, it reduces the requirements by one.
https://www.asus.com/us/Laptops/ZenBook-Pro-Duo-UX581GV/
Two screens, lots of power (high clock speed in addition to core count), I like the keyboard. Gorgeous main screen. The second screen is treated just like a normal extra monitor, not limited by proprietary software, has same dpi as main screen.
Completely unrelated, but I highly recommend checking out Civilization 5 with the Vox Populi mod. On the whole, Civ 5 is slightly worse than Civ 4, but the Vox Populi complete overhaul mod makes the game the best Civilization game I have ever played. I still play it on the regular.
For 1 (Civ) A HP ZBook. You said money and size no object, and this is it. Alternatively a Thinkpad P series or Dell Precision.
For 2: Get a USB webcam and a tripod. The Logitech C920 is ancient but still better than almost any built-in webcam and mic. And, you can unplug it when you don’t want the possibility of being snooped on.
Edit: Two tripods. On the second one which is behind your screen, put a cheap 10×8 inch 192-LED lighting panel, available from any good photographic supplier. The batteries and charger for the panel will probably cost more than the panel itself. Get an extra set of batteries. If you want to take your Zooms from terrible to OK, this is a sine qua non.
And, for the love of Moloch, use a wired ethernet connection! 100M USB 2.0 ethernet adapters are cheap as chips, and wildly better than any “1300 megabit” wi-fi connection. 10-metre (um, 33-foot) ethernet patch cables cost peanuts too.
For 3. You want a Lenovo Thinkpad T series. The T420 was the last one that had the really good keyboard, but even the T490 is still better than nearly anything. I’m not sure if the P series has the same keyboard; I can’t afford one, and the local big box store doesn’t have ’em.
(Edit 2: There’s a reason Thinkpad Ts have the best support in the open source community — Linux, FreeBSD, etc — and it sure as heck isn’t the screen or the touchpad. Thinkpad Ts are one of the few things that lawyers and hackers find common ground on.)
XPS are thermally limited and underpowered to start with so won’t play Civ the way you want it. Their main selling point is thin and light. Surface Book 3 is … OK: nice screen, good touch panel (which you didn’t mention), but not very fast for the price. The keyboard isn’t for me.
If any of you enjoy traditional Gaelic singing combined with electronica and drone metal: https://fosmetal.bandcamp.com/
I want to raise a bunch of geniuses, taking some inspiration from Laszlo Polgar (though not focusing on chess). Polgar planned to raise six, which sounds decent to me. One thing I don’t share is Polgar’s belief that any healthy person is equally capable of genius; I intend to stack the genetic deck in my favor as much as possible.
I want the genetic mother to have the following traits: (a) extremely high intelligence, (b) no serious mental, personality, or health problems. For (a), an IQ of at least 160 or performance on academic contests at the 99.99th percentile would be good. For (b), something like “mild ADD or occasional depression” is fine as long as it hasn’t severely impacted your career or social life. (For the curious, I do meet the above criteria myself, although for the later children I would consider getting the sperm of someone like Terence Tao.)
If you have the above traits and are willing to give me your eggs, I’d like to hear from you. If you don’t have the above traits but think you would happy to help me raise children according to this plan, whose eggs come from someone else, I’d like to hear from you. If you have the above traits *and* want to raise children with me, I’d definitely like to hear from you. I’m a programmer at a medium-sized Silicon Valley tech company, so finances should be ok.
I can be reached at polgarity at protonmail dot com.
Amused at the idea of you furtively acquiring genius sperm. I suspect this resource is not well defended!
Because children are public goods and your idea is so monstrously politically incorrect, it plausibly has very high leverage. In fact, given the heritability of cognitive ability one of the most high-leverage things a genius can do is reproduce.
It is a shame you are not a rich female genius, as then your success would be assured.
If you are really serious about this, you may be able to make trades others would not. The sisters and daughters of intellectual high performers, even those who are conventionally unattractive, should also be interesting to you, even if they are not as high performers themselves.
Regardless, I wish you luck founding your dynasty and hope you find love in the bargain.
I don’t meet the above criteria, just popping in to question why someone would take inspiration from Laszlo Polgar but also completely miss his thesis that nurture overpowers nature. Are you trying to disprove him? Are you after the prestige, so that someone 50 years from now will write a blog post about you? This seems not very well thought out, and I have to wonder if this is a thinly veiled attempt at circumventing the normal frustrations of online dating.
I think “explicitly noting and rejecting his thesis” is different from “missing his thesis”. Polgar also repeatedly cites Glenn Doman (e.g. “Glenn Doman, one of the most well known scientists who study genius, has a similar opinion: according to him every child before the age of 3 can be raised to be a genius”); if you look up Doman’s practice at https://en.wikipedia.org/wiki/The_Institutes_for_the_Achievement_of_Human_Potential , you see “multiple studies have found the therapy ineffective”.
If I were trying to disprove the general nurture thesis (i.e. that it’s all nurture; I believe it is a combination), I would just collect the studies that have already been done about life outcomes of twins separated at birth. If one wanted to disprove “Polgar’s nurturing makes nature irrelevant” with this experiment, having a control group of children with average biological parents would be essential; surely you don’t think a reader of this blog would make such an elementary mistake in study design.
To understand my motive, I recommend reading “Exceptionally Gifted Children” by Miraca Gross. She did a long-term study of 60 kids with 160+ IQ in Australia and wrote a detailed book about 15 of them. The majority of them had to spend most of their school career attending classes designed for their normal age-peers; obviously this wasted their time and was painfully boring, but perhaps more importantly, the severe mismatch with their peers made it a very inappropriate social group from which to try to make friends, leading to isolation and, probably (it’s difficult to quantify this, but some efforts were made), stunted social development whose effects might last a lifetime.
Contrariwise, a few of the kids were allowed to skip multiple grades. One of them, codenamed “Adrian Seng”, had probably the best educational program, letting him go through two grades per year in some subjects; at age 7.5 he was taking math in high school and some other subjects in elementary school (and was “extremely popular with both his elementary school and high school friends”); by age 9 he spent 3/4 of his time in high school and 1/4 at university; at age 14 he was fully in university. He was the only kid of the 15 to say that he has been able to perform at the limit of his abilities in his schooling. He became a Fields medalist and is probably in the top 100 mathematicians in the world, top 10 according to some people. I mentioned Terence Tao earlier.
The kid with the second-best educational program of those 15, “Chris Otway”, did a decent amount of skipping and could have entered university at 14, but chose to stay in high school for two more years and took a broader range of classes. As of when Gross’s book was published, he was working for a “worldwide strategy consultancy”, where he “finds that his mathematical talents and his study in economics are both highly valuable” and was promoted within a year. I think strategy consulting has a bad reputationβprobably in part because the customers are, by assumption, not really qualified to evaluate the advice given (unless they hire multiple firms and compare recommendations), allowing bad actors to flourishβbut at least he seems a highly capable professional.
An earlier work, “Children Above 180 IQ Stanford-Binet” by Leta Hollingworth, gave detailed information about 12 such children. One of them, “Child L”, attended Hollingworth’s own specialized Speyer school, and seems to have become a fairly good mathematician with his own Wikipedia page and an algebraic structure named after himβlook up Murray Gerstenhaber.
My point is, if you take kids with these kinds of IQs and give them a proper education, there is reason to expect very good results. Not from every one of themβsome will be drawn to fields that aren’t especially productive (no offense to the pure mathematicians above), others won’t have high motivation (trait “Conscientiousness” or whatever) or won’t work out for other reasonsβbut with six or so kids, I think there’s a good chance that at least one of them will end up creating something very valuable in tech or biotech or elsewhere. (And merely becoming an excellent employee of some company is not worthless either; society needs those.)
And even those who don’t contribute something great in themselvesβif they carry on my project and do the same thing with the next generation, then you have a bunch more kids with the potential for greatness. I have played way too many strategy games to not have a profound appreciation for the power of exponential growth.
Regarding me, I think my own motivation isn’t that high, and I have ingrained habits of “screw you, I’ll do my own thing” and other difficulties with socialization and authority that I would trace back to a childhood in which I spent most of my day bored, around kids I couldn’t relate to (with a couple of exceptions, who unfortunately were usually not in my classes), at the behest of adults who claimed to have my best interests at heart but mostly ignored my protests. I think I’m improving, but it has certainly handicapped my careerβI threw away some opportunities early onβand some difficulties linger, and while I think I’m ok, I’m not sure if I’ll ever get great at working with others who aren’t technical geniuses or close to it (fortunately, my company does have a reasonable number of those). I also seem unable to maintain a normal sleep schedule without sleep deprivation, which is less bad for programmers but still an issue. (I refuse to use caffeine or other stimulants.)
Anyway, I think I am likely to remain merely a highly capable programmer working at a company, with some chance of making something great in CS on my own, and maybe some chance of being a technical cofounder at a company that does something great. It is also possible that simply being a “sane and thoughtful person” in the right place at the right time, to override a disastrous decision or find a great opportunity that others overlooked, will outweigh years of normal work. All that aside, I think that the project of raising geniuses is pretty likely to be more valuable, in the very long term, than everything else I might do. (If people figure out how to modify genes to make everyone geniuses with no negative side effects, then maybe my efforts are redundant, but that would be a “nice problem to have”.)
One reason to have a lot of kids is to give them something of a peer group that way. I also intend to settle near other parents whose kids are likely to be of similar caliber, and encourage them to make friends. The best peers for extremely gifted kids are equally gifted kids of the same age; second best is moderately gifted kids a couple of years older.
About recognition, it would be nice if I achieved my aims and my friends and family appreciated me for it, but I don’t think I’d want to be famous for it (Polgar himself complained about getting criticism from people who assumed he was slave-driving his kids). What I would really like is if a bunch of geniuses were inspired by my project to do the same thing themselves, and then, years later, I and my children met a bunch of well-raised, well-socialized geniuses and we delighted in each other’s company, and a few of us went on to fix some of the worst problems facing the world while others were mere solid contributors or made some nice games. Actually, because of the “peer group” thing, it would be better if I met them much earlier. (Email me if you’re considering doing your own project; I currently assume I’ll stay in the Bay Area.) But best would be if there were enough families doing this that I’d have enough for the early peer group and for meeting a bunch more as adults!
Are you considering using standard egg donor agencies? Having been through this process myself, it is pretty easy to find candidates with SAT/ACT scores that put them in the 130-140 IQ range (very easy if you have no other constraints, eg appearance or other desires). Meaning that going through a few hundred profiles over a day or two across a few agencies, you’ll find 4 or 5 candidates like this. This would cost you perhaps $15-$30K total to pay both the donor agency and the donor (plus further fees for the egg retrieval itself, and then more for IVF after, but both of these seem to be part of your existing plan?). Getting a higher level of estimated IQ would probably require hiring a specialist donor agency that has access to pools of women at the very top schools – this might cost you more like $50-70K total. Definitely advantages to finding the donor yourself – you don’t rely on the donor agency being honest and can vet the candidate yourself (otherwise everything is anonymous and you get pictures and a text description from the agency, maybe a video). And obviously cheaper.
I had looked at a few egg donor websites some years ago. A couple of them wanted contact information or something before they’d let me see any profiles. The one without that requirement seemed mainly to focus on photos of the donors and I think the closest it came to indicating intelligence was mentioning level of education and current profession. I’ve also looked into sperm donation and I think the above applied to them too; physical health and attributes (height in particular) were the main things it asked about other than educational attainment. I figured that must be what their customer base was asking for. But it would make sense for there to be other, more specialized agencies.
Compared to the cost of raising a kid overall, something in the low tens of thousands per kid seems pretty reasonable. The $50-70k makes me pause, but if it’s a one-time fee to find a suitable source for all the kids, that could be fine.
To even have 10,000 different scores on a test, there must be log(10,000) = 14 different questions (assuming that each question has only two categories of answers). That’s with the results being a binary tree. For there to be 10,000 different results that are monotonic with the number of questions answered, there must be at least 10,000 questions. For those buckets to be statistically rigorous, there must be about a billion. So the very fact that you think you have a valid basis for believing that you have an IQ in the 99.99th percentile is itself strong evidence that you do not in fact have an IQ in the 99.99th percentile.
Regression to the mean is stronger the further you go from the mean, so even if you could identify a woman at the 99.99th percentile, this would provide little benefit over finding someone at the 99th.
Excluding more than 99.99% of the population is going to come across as conceited, even (perhaps especially) if you have chosen criteria that you have met.
There are many attributes that benefit society more than intelligence does.
I don’t think that analysis works.
Suppose we have a question to which there is provably a right answer. Our test consists of giving someone a minute to answer the question. We give the test to a hundred thousand people, and a hundred of them succeed. That’s a single question, and the test gives evidence that someone who answers it correctly is at about the 99.9 percentile in the particular skill it is testing.
You need some additional assumptions that I don’t think you have stated.
There are some assumptions that are implicit, such as that the test is not focused particularly on the 99.99th percentile bucket. (I did explicitly state the assumption that the test has 10,000 buckets, and it can be inferred that each is being treated equally). Even if there’s a test that focuses specifically on the 99.99th percentile, it would take about 100,000 test takers to make it valid. So we need some competition that focuses specifically on the 99.99th percentile and that 100,000 people have taken. I’m rather skeptical that such a test exists.
IQ tests have been given to many more than 100,000 people, and there is no reason to assume that what percentile you are in is equal to the fraction of questions you got right, which I think was your implicit assumption.
One definition of IQ, I think still used for children, is ratio of mental age to physical age. One could give a five year old a test designed for ten year olds, observe that he did as well as the average ten year old, and assign him an IQ of 200. By examining the distribution of such IQ’s one might conclude that that was at the 99.99 percentile.
For adults, one could give a test sufficiently hard that nobody got every answer right, observe the distribution of scores, and conclude that a score of, say, 99% put someone at the 99.99 percentile.
Also, what the poster you were responded to said he was looking for was not an IQ at the 99.99 percentile but
Do you think either of those is unrealistic?
My version was “someone I can discuss ideas with without feeling as though I need a translator.” Producing children of whom the same was true was a pleasant side effect.
This is possibly the spergiest thing I’ve ever read here.
I am going to take this as a compliment.
For what it’s worth, I grew up in what was surely a high IQ family, I expect the same is true of my present family, and I think you are doing it wrong.
Ah, interesting. Could you be more specific about what I’m doing wrong?
A fair question. I’ll try to answer it.
The way to produce people who are very smart and functional isn’t by having a project to produce six of them. It’s by creating the sort of family in which each child grows up feeling loved and seeing rational behavior as the normal way of interacting. Having parents who see you as their project is likely to work less well than having parents who love you and enjoy having you as part of their family.
Suppose you get eggs from a very smart donor and marry someone substantially less smart who, at the point when she marries you, says, perhaps truthfully, that she is happy to go along with your project. Bringing up six kids is a lot of work. Dealing with husband and kids who are substantially smarter than you are, and know it, is a potential strain. You might get lucky and find a woman who can function well in that situation, but it isn’t the way to bet.
You are better off finding a very smart woman who loves you and you love and who wants to bear and help rear your children. If you succeed in creating a happy and functional family, the odds are good that at least some of your children will end up smart and functional.
The version where the mother of the children is the one who helps you bring them up works better than the other version, but mutual commitment to a project you both thought you approved of in your twenties (I’m guessing β you might be older) isn’t much of a guarantee of the next thirty years of a relationship.
And would you want to be brought up by someone who saw you primarily as his project, to whose success he had committed himself? What happens to your relationship to a kid who doesn’t seem to be turning out as the genius you designed him to be?
I hope that helps.
P.S. The one member of my current family who has read what I just wrote thinks I am being uncharitable, that your project could easily turn out as the sort of family I am recommending.
For people who like non-fiction audio books, two of mine are now up on Audible, Hidden Order: The Economics of Everyday Life and The Machinery of Freedom: Guide to a Radical Capitalism, 3d edition.
My first novel, Harald, published by Baen, is also up as an audiobook.
All three books are read by me.
Two years in the making, this tiny, 143-page book has been meticulously engineered to evoke a sense of hope during the apocalypse. It will inspire you to change your life and change the world. Anarchy, attachment, cars, economics, entrepreneurship, entropy, ethics, grief, guilt, guns, kindness, self-reliance, seventies nostalgia, and stoicism. Awaited: Nonfictional Delusion, by John Binns.
https://youareawaited.com/awaited-nonfictional-delusion
A friend and I are brainstorming a project called “Renewables Are Not Enough.” It’s meant to argue against the techno-optimism around renewable energy in climate discourse. It should be very straightforward to anyone who has spent some time on the issue, but I worry that there is not enough attention and discussion on complementary solutions to renewable energy. In particular, we think there isn’t enough attention to reducing consumption and limits to growth.
We’d appreciate feedback from anyone interested or involved in this area. Here is a one-pager proposal which includes a summary and three questions we could use feedback on. Take a look at this Google Doc and leave comments there: https://docs.google.com/document/d/106L9VQn8DQLOmjiicn2SVTpXDbpgAuoBRwE7ZJVTOPY/edit
Single, 32 year old man looking for a long-term partner in Madison, Wisconsin.
You can look up various pictures and things I’ve said online pretty easily by googling “Drethelin”, my Instagram has a lot of pictures of me as well as some of my hobbies. I read a lot, watch a lot of stuff, and talk to people online a lot. You can email me at the obvious address or catch me on whatever social media website is convenient.
I’m not looking, but it might be helpful for those who are if you indicate whether you have any gender preferences for a long term partner.
[I am Ludovico, Canyon Fern’s human assistant. I am looking for work in the United States: part-time or full-time, remote or on-site, ideally in technical writing, editing, translation, or software development.
I hold a BA in Linguistics. My native language is English, but I speak and read Mandarin Chinese as a second language, and I can program computers (mainly in Python.) I’ve worked in computational-linguistics research and Chinese-to-English translation. On my own time, I’ve created websites and text-adventure/IF videogames. I also have a stupendously strong sense of humor.
I’ll consider damn near anything, and I’d be delighted to assist with small projects. You can reach me at contact.canyon.fern@gmail.com, after which I will contact you under my True Name.]
My name is Roddy Boyd and I am the founder and editor of the Foundation for Financial Journalism. It is a 501(c)(3) whose mission is to provide long-form, accountability-oriented investigative reporting on the capital markets. We are 100% donor-supported, conflict-free, don’t accept ads or sponsorships and above all, don’t profit from market movements (i.e. we don’t short stock.)
I argue that our reporting has broken some of the bigger business stories of the past eight years, and has been responsible for sending more than a few to jail.
For what it’s worth, our reporting is remarkably crisp because it is tightly edited, and every word is extensively reviewed by an experienced team of libel lawyers.
I am hoping to grow our readership base and, to be utterly frank, our donor base. I have fallen in love with SSC and figured now was a time to stop lurking and present myself. My hope is that SSC readers will take to our reporting in Wall St crooks et al, share it with their friends and maybe hit our PayPal.
Anyhow, our website is ffj-online.org. Just compare FFJ to wherever else you get your business news, and I think you’ll like us.
This relevant for my interests :). By the way the link points to https://slatestarcodex.com/2020/05/07/classified-thread-8/ffj-online.org should surely be just https://ffj-online.org/
I fixed the link.
Thanks so very much Scott. Your site has given me so much.
Oops, I’m sorry.
It’s an interesting site. I know (loosely) some of the folks you’ve referenced in your writing.
You said you’re focused on long-form bigger stories, but it seems that posting on a more regular schedule than say, intermittently with sometimes months between posts, would likely encourage more consistent return traffic. Even weekly would help your cause. At least if you expect people to regularly check back at the site to see if there is new reporting.
Otherwise, you might consider periodically subletting some of your articles to other sites, such as the WSJ, in order to increase discoverability.
Your ideas have a good deal of merit, and I thank you for reaching out. I plan on increasing the frequency of posting (without losing the depth and accuracy that FFJs readers find meaningful.)
I’m a data scientist in training, looking for interesting projects, paid or volunteer, to get my hands dirty. I’m not terribly concerned with pay at this point, more with opportunities to learn stuff and develop a portfolio. Education-wise, I’m 1/4 of the way through a master’s in DS, after two years of doing data analyst stuff in finance.
Email: patrick24601 at gmail dotcom
I am writing a blog and/or starting a transhumanist church!
https://transhumanist.church/
yes hello how do I join
Howdy. editor, tutor, campaign consultant, at your service.
Editing/Proof-reading:
I have an econ degree from an Ivy-league, where I was the sharpest-eyed among the editorial staff of a biweekly newspaper, and I’ve done paid work as a journalist or editorial support for three different respectable institutional publications. I also have a law degree and know my way around every punctuation mark. I know how to write defensively, whether by being so dense as to be indecipherable or so carefully worded and reliant upon authority as to feel compelling. I’ve more than a year’s experience working as a technical writer for a Fortune 500 biotech firm. If you need to know something is typo-free, you can’t find better help than me. Prices may vary by the job and the client, but generally: $35/hr for proof-reading only. If you prefer to pay per word or page, I can work with that. $60/hr for more thorough editing work including making editorial suggestions and fact-checking. $100/hr for technical writing, if the project is small and the delivery date flexible.
Price may be higher for projects with aggressive time-tables or where I have to commit to being available for a long period.
Campaign Strategy/Management Consulting:
I’ve got 8 cycles of campaign management and two years running a side-gig as a consultant-for-hire working with local political candidates. As a consultant, mostly I’ve been hired by campaigns with a full-time campaign manager and/or a full-time full-service consultant, where my role has been as a second set of eyes or an extra source of ideas, particularly as someone with a good strategic understanding and who is completely independent of state political machines. I have served as a full-service consultant once, and have good relationships with multiple vendors for every campaign need. I really feel passionately about this work, and am available for this work for a candidate I like at well below my market rate for other skills. @JRM is a longstanding commenter here at SSC that I have worked with, who found me through a previous classified thread, who I think will recommend me.
Tutor: if you would benefit from help planning your attack on a standardized test, I’ve taken many of them, and led study-groups with high success rates (including one A.P. study group in high school where 100% of my students got 5s… still very pleased with that.) I’ve helped people plan for AP, ACT, SAT I, SAT II, GMAT, GRE, MCAT, and LSAT. Not all tests are equally teachable; e.g. for most test-takers, the LSAT is much harder to meaningfully improve on than the MCAT, but all tests have it in common that there is some gain available from understanding how the testers aim to achieve a ‘hard’ test. (Some rely much more on time pressure, others test for whether you’ve got a variety of good quick tricks for solving impossible-looking math problems, others are closer to raw IQ-type tests.) I won’t promise you the moon or charge you for things I don’t think will help. I will do everything I can to give you a study plan that matches your available time and goals. $40/hr, generally capped at $160 per plan, can price to suit your budget.
I’m at Kh_mec_k@g_mail.c_om except remove the underscores.
Looking for remote work in ecommerce, product management or analytics.
After spending the last few years working on my own ecom business, Covid-19 drove us under. Sales dropped by >50% in March and the outlook for a return to baseline anytime soon isnβt great. So now Iβm looking for work again.
I spent almost 10 years at a small cap ecom firm doing everything from pricing strategy to UX design, have done web/online strategy consulting for orgs like Stanford Law and MIT Technology Review and spoken at a few conferences about consumer psychology and product development. If youβre hiring or know anyone who is, would love to hear from you. Open to contract work or a full time gig for the right company.
Thanks,
Chris
chris@databoxdigital.com
linkedin
By day, I’m a law student (just about to finish). By night, I make text-based computer games. Here’s my gameography in order of how good they are:
Skybreak!
Link
Manual:https://ifarchive.org/if-archive/games/competition2019/Skybreak/Walkthrough/The_Manual_.pdf
wiki page: https://ifdb.tads.org/viewgame?id=b21j5feykymttixm
Skybreak! is a very large choice-based role-playing game where you fly around an elaborate science-fantasy universe, making one decision on each world. It won 12th place in last year’s IFComp as well as the Golden Bannana of Discord (for the game with the highest standard deviation among judge scores)
I Summon Thee!
Link
A piece of speed IF made for a lockdown-themed comp. Every game was built around the theme of escape. This is a semi-standard parser game that basically plays like Arkham Horror in reverse. Sparse, but loads of fun. It’s also a good example of what you can do with ADRIFT over two 8-hour workdays.
Tingalan
Link
Manual: https://drive.google.com/file/d/0B4rZaAbhHw61UnhQdDRSSm9iaXc/view
A survival/horror game where you collect treasure from the depths of a fairy-forest. Tingalan is…very hard, and the commands are a bit picky (it was my first game) however, people on this site have enjoyed it.
The Dead of Winter
Link
Another speed IF, this one made over 24 hours in two 12-hour shifts. A very dark military-themed survival game. a bit buggy simply because of the time constraints.
Six Silver Bullets
Link: http://www.adrift.co/game/1530
Link to ADRIFT runner (which you need to play): http://www.adrift.co/download
This was an entry for IFComp 2018. A parser with light puzzle elements and a lot of intrigue. Sadly, it’s not playable online, so you’ll need to download ADRIFT (which will only run on windows)
I am a retired business consultant who specialized in Organization Change Management, Organization Development and Executive Coaching. My first career was as a psychotherapist in a psychiatric hospital, but left that position after five years to work for a consulting firm. I am looking to return to my psychotherapy roots and have just completed my fifth year of a six year program in Body-Centered Psychotherapy. Because I am no longer licensed I cannot say I am a psychotherapist and cannot take insurance. I am interested in taking on additional clients. I receive supervision from the Hartford Family Institute where I study. I use tele-health applications for therapy sessions.
If you are interested in low cost, high quality therapy, email me at smbsmb4@icloud.com.
I’m a front-end web developer with 5 years of experience, looking for remote work. My background is in visual design, and I have some training in back-end development, but I’m strongest at making complex things easy to work with by visualizing them and building good interfaces for them. I live in Texas but I’m open to relocation for the right opportunity. Here’s some of my work: https://jamesharris.design
I would greatly appreciate any leads or advice for my job hunt.
I’m in my late 20s, with a bachelor’s degree in computer science and mathematics from a university that excels at those, and a master’s in privacy I obtained from a program at the same school. After finishing undergrad, I worked in software engineering for several years at Microsoft and Booz Allen, before concluding that software engineering wasn’t the sort of work I wanted to be doing and going to graduate school. Since getting my master’s, however, I’ve been on a fairly fruitless search for full-time work, primarily in the privacy domain.
While I’ve had my fill of software engineering, I’m still a solid programmer (though I have very little experience with webdev or database engineering), and I’m excellent at discrete math. Outside of technical skills, I’m a talented writer and speaker who can pick up new domains quickly, as when I wrote the report for and presented at a panel on child online protection in Africa. I enjoy policy and teaching, and when including my part-time work, my favorite jobs have been in those domains. I do get somewhat scared of bureaucracy, which has dissuaded me from seeking government work or becoming a high school teacher. I could see myself getting a doctorate or JD at some point in the future, but I would prefer not to just yet.
Some projects of mine:
* I created a puzzle hunt SCP, along with quite a few other well-regarded articles.
* I designed and implemented the Better Bosses mod for Slay the Spire.
* I’ve developed several original magic tricks, including one published in the trade magazine of the International Brotherhood of Magicians.
* I conducted a legitimate parapsychology research project, inspired by our host’s The Control Group is Out of Control… and actually got positive results. (I’ve shown my report on this to other folks in the SSC community, and our conclusion is that it’s genuinely a good study, just with a weird result that’s probably chance.)
* With just a couple weeks preparation, I travelled to Ghana and presented before representatives from dozens of African nations as a UN expert on best practices for child online protection, then wrote a thorough report on the topic.
* I grow copper and bismuth crystals at home.
I’m based out of the DC area, but I’m entirely willing to relocate. If anyone knows of any roles I might be suited for, or has more general advice, that would be very welcome. I’m happy to answer any clarifying questions. You can contact me at euclid11 [at] gmail.com.
When are you going to finish SCP-2212?
Oh god, here too?
I have an outline and a partial script for the final encounter/puzzle. There was someone I was working with on an interface for it, but I haven’t contacted them in a long while. I would still need to finish the script, and write a few modified versions of the article representing the end states. As more of a reach goal, I’d like a CSS that distorts the SCP site to capture how SCP-2212 is messing with the Foundation. There’s a couple other alterations and additions I’d like to make, but those are more minor. There’s other, non-SCP projects that I’ve worked on that have meant I haven’t gotten around to finishing 2212, but I promise you, I do want it done.
I make music and do sound design for short film, indie video games, and visual art projects. If you have a sizzle-reel or a draft of your project and would like some audio for it, get in touch – you’ll be glad you did!
Email in ROT-13 (https://rot13.com/) below:
ivyrzrqvn@cebgbaznvy.pbz
TL;DR: Looking for a job in the Bay Area or possibly in DC. I’m an aerospace engineer at a tiny 10-person company called Spacequest; we designed, built, launched, and currently operate a pair of 3U cubesats in low earth orbit. I enjoy my current job, and I would be excited to continue working on new projects in the space industry. But separate from my specialized aerospace knowledge, I also think I have some underutilized potential as a general researcher and information-synthesizer — over this past year I’ve really loved working on a remote team with Phillip Tetlock and 15 teammates to clarify key questions & disagreements, make fermi-style estimates of crucial variables, and construct testable counterfactual forecasts. I’ve loved it and I really want to do more work like this — but where should I be applying?? My current dream job might be Research Associate at OpenPhil, but if they don’t take me, what other places do this kind of information-synthesis work?
Aerospace experience:
β’ I handled several aspects of mission design on our two 3U Cubesats, Brio and Thea: thermal & power management, spacecraft operating modes, attitude determination & control systems. Wrote and tested embedded C code controlling ADCS operation (aka controlling where the satellite points in space, using sensors and little spinning wheels and electromagnets), antenna/panel deployment sequence, initial ground contact procedure, and recovery/failsafe features.
β’ Once we were in orbit, I calibrated the satellites’ ADCS system, plus helped create software fixes to work around component failures and solve other unexpected problems. These days I’m mostly writing python and bash scripts to automate recurring aspects of satellite operation including telemetry analysis, command scheduling, and hosted payload operation.
β’ On a separate project while I was working on my Masters degree in 2016-2017, I designed a unique ADCS strategy for the “C.U. Earth Escape Explorer”, a 6U deep-space cubesat mission, using solar radiation pressure instead of thrusters to desaturate reaction wheels. I wrote a detailed Matlab simulation to predict torque buildup during the mission and plan desaturation maneuvers, then produced analysis and documentation to demonstrate the idea’s feasibility to a panel of NASA judges, contributing to CU-E3βs selection for launch on the first flight of SLS (whenever that happens, lol).
β’ Honestly, my coding skills (C, Python, Matlab, some other stuff) are serviceable, but I don’t get super excited about writing code for its own sake. Even as an aerospace engineer, my main value-added has been that I’m a huge space nerd and I know lots of detail about how different spacecraft subsystems interact — if you want to design a mission and you’re wondering how to think about orbital dynamics, ADCS, thermal considerations, power budgets, and etc, I’m your guy.
Superforecasting:
β’ I’ve participated in three rounds of IARPA’s “FOCUS” competition about predicting “counterfactual” outcomes of complex simulations; I’m currently one of 3 leaders on the upcoming final round. (We get to see the “original” run of the simulation, then we get told about a key change, then we have to forecast the alternate history.) Simulations have included cellular automata, chess matches between AI opponents, the acclaimed sleep-destroying videogame “Civilization V”, and the “SIR” pandemic epidemiology models that we’ve all become so familiar with these days. In three rounds and with six teams each round, my team has placed 1st, 2nd, then 1st. Prof. Tetlock talks about the tournament in this 80,000 Hours podcast, which is where I first found out about it.
Some of the kinds of optimizey, information-synthesizey projects I like to work on in my spare time, which give me the feeling I should maybe pivot my career more towards this type of work:
β’ Trying to estimate the net-present-value of all the wages my wife and I will likely earn in our future careers, plus social security benefits, real estate, etc, to create a picture of how the components of our overall wealth will change over our lives, and using that picture to inform the portfolio allocation of our traditional financial investments.
β’ Creating jury-rigged quantified-self system using some python scripts that run once a day, pulling data from Fitbit, Rescuetime, and a few other services, then sends me an email with customized graphs describing how well or poorly I’ve been sticking to various habits and goals.
β’ Starting a blog project (not yet launched) together with my wife, who is a quantitative ecologist, where we’re hoping to explore the intersection of longtermist effective-altruism with ecology and the environmentalist movement.
Final thoughts:
I feel like I’ve found a type of work where I’m happiest and easily motivated to work hard, because I’m working at improving a fundamental skill. I’ve also loved the intense feeling of team collaboration (even though it’s all been remote!) during the IARPA competition. The atmosphere of working together with a bunch of smart people to think clearly about messy problems, in an environment with real stakes, is very compelling — before this, I thought I didn’t like team projects!! I have the sense that this skill is at least somewhat valuable as career capital, but I don’t know where exactly I should be looking for an alternative career path. EA research definitely fits the bill. Sometimes people talk about certain unusual hedge funds and other financial firms having a similar high-stakes intellectual culture? (But they also sound very competitive, and my on-paper experience is all about engineering.) Here in DC there are lots of assorted think tanks and people do lots of “consulting” — does these options exist in the Bay Area as well? Finally, there are technical but broad and exploratory research labs at places like Xeroc PARC, Facebook Reality Labs, and similar outfits at other big tech firms. But it’s a big world out there, and I feel like I’m probably missing a lot. If you can help come up with ideas of what career directions might suit me, I will be eternally grateful to you!
In fact, I am already grateful to you for reading this very long comment! If you want to talk even more, you can email me at “jackson.wxyz [at] gmail (dot) com”.
Management consulting (think McKinsey and the like) has many elements that fit what you describe. It also puts a large emphasis on getting stuff done in organizations, which may or may not excite you.
Now is a tough time to be looking for a job in general, but Iβd suggest looking into the field.
I’ve been looking for something like Duolingo for identifying logical fallacies and rhetoric, but haven’t come across quite what I want, so I created this survey as a proof of concept:
https://www.surveymonkey.com/r/X78WJW3
Please give it 5 minutes, and if you like the idea let me know. Thanks!
Done. That was fun!
Tried it, was fun!
One piece of feedback, if you care:
I ended up having to use process of elimination on question 6:
What tripped me up was the “She’s way more knowledgeable,” since it’s not clear where knowledge that the PM is more knowledgeable came from. If, in this universe, your PM is a Nobel Laureate, this statement conceivably be accurate. Something like, “She always talks about how smart she is” or “She couldn’t have gotten elected if she wasn’t smart” or something would make it more clear that you’re angling towards this being circular reasoning.
Cool! I’m in the final stages of writing a book about optionalityβsomething along the lines of ‘Nassim Taleb meets Mr Money Mustache’βand would love to enlist some help.
1. Illustrations
Looking to commission roughly a dozen line drawings. For reference, I love Edward Gorey, and John Tenniel (the classic Alice in Wonderland illustrations) but am open to ideas. I don’t have a huge budget for this, so if you (or someone you know) is still on the amateurish end of the spectrum, that’s fineβit would be neat to give someone their first commission, especially if they’re an SSCer.
Please email some examples of your work and a rough sense of your pricing to richard dot meadows at thedeepdish dot org
2. Alpha readers
I was thinking of doing a reverse auction for this, but it’s probably better to just ask if anyone’s intrinsically interested. Again, the topic is optionality: why it’s becoming more valuable, how to get more of it, what are the limitations, when to deliberately sacrifice options.
I’m looking for generally smart people who might be able to point out mistakes, and give structural feedback on what works and what doesn’t (not typos or line editing, although that’s fine if you’re so inclined).
The book includes a bunch of rationalist-adjacent concepts, but it’s written for the general public. To get a sense of whether this is likely to be of any interest to you, here’s the (draft) table of contents.
You could read one part of the book, or the whole thing if you’re really keen. Again, please email me on the address above if you’re interested.
Thanks!
Sent you an emailβI am tentatively interested in reading your book. Do you have a one paragraph explanation of what optionality is?
Cool thanks, will reply you there.
Optionality = the right, but not the obligation, to take a given action.
A high-quality option displays an asymmetry between risk and reward: you pay a small, fixed downside cost to get exposure to a large or unbounded upside.
I think I’m a fit for your alpha-reader need. You can see more about me in my post down-thread.
Send me an email with what you’re thinking in terms of compensation; I’m certainly interested. I have to prioritize paid work, but I could see the final product of what you’re working on being something I would read for fun, and I did recently find myself with loads of time on my hands and a real need to find work at whatever price, so make an offer.
I also think my sister is a likely match for your Edward Gorey needs, and I understand she’s recently found she has some time available.
This evil herbivore is from several years ago: https://m.imagekind.com/evil-herbivore_art?IMID=78879c71-ba13-4394-9c68-d578c164adc5
This redbubble site has some of her work also: https://www.redbubble.com/people/dmuck7/shop. I’d say look at the ‘celtic lobster’ if you want to get a feel for her ability to evoke very specific and complicated animals/feelings with very simple lines. Her main artists’ page is down right now because Disney noticed it riffs on Winnie-the-Pooh, and, uh, don’t mess with the Mouse, apparently. Mention if you are interested in her work in an email to me, and I’ll connect you.
Thanks Kelley. Your skillset looks perfect but I have enough volunteers for now that I’m probably set – if I decide to hire someone to do thorough line editing I will def keep you in mind.
I forwarded this on to a friend who’s an illustrator.
Thanks!
Sent you an email.
Thanks, will reply you there.
How many words is the book?
Over 100,000, which is why I’m happy if alpha readers choose one section (say ~20,000 words, which might be an hour or two of reading).
Hi! If you have a software project in mind and are looking for an engineer to build it, let me know. I’m Austin, an ex-Googler who now works on a freelance basis, with a pretty wide background (a lot of web frontend recently, but have also done backend, ML pipelines, VR, mobile). I’m not attached to any particular technology though; my focus is just on building stuff quickly and reliably, with high attention to detail.
I’ve previously worked with onyomi based on a post on Classified Thread 6, which has went quite smoothly (hopefully onyomi agrees!), and I’d love to find other clients who frequent SSC. Reach out to me at akrolsmir@gmail.com.
The Society of the Free & Easy is a free & open source, self-guided, peer-supported technique for strengthening virtues.
Virtues are habitual characteristics of thinking and behaving that tend toward a successful, thriving, and beneficial life. There are intellectual virtues (like rationality), virtues of personal character (like courage), and virtues concerning social interactions (like empathy).
One influential theory about virtues holds that, like other skills, they can be acquired through practice and become habitual over time. The Society of the Free and Easy is a method for integrating this sort of practice into your life. By working together, we improve ourselves, help those around us improve, and also improve the methods that make all that happen.
If you think you might be interested, I can give you more details. Or visit our wiki for the whole brain dump.
I checked out your wiki and I think this is a cool idea. I’m a big fan of Aristotle and of virtue more generally. One thing in particular I noticed, on the page for Christian Virtues:
This reads as advice against cultivating Christian virtue, on the grounds that it will not help you enjoy life and may hurt. That seems straightforward enough from a theoretical and theological perspective. However, empirically, there is a lot of data (example) indicating that religious people are more happy. The data is less clear than we might want it to be, and I think it’s reasonable to harbor doubts or alternative speculation about the mechanism and limits of the correlation are. However, for a society that values empirical results, the presumption should probably be in favor of Christian virtue contributing to a happy life, not against. The conclusion could be made more helpful by pointing out that, theory or theology aside, practicing Christians have no shortage of happiness, and usually credit their religion for it.
Anyway, having dismounted from that hobby-horse, like I said above I think it’s a cool idea and I might be interested in joining. I’m a little unsure that an online group could help overmuch for cultivating virtue, though. Could you say a little more about how it works?
Second question first: I don’t think of us as “an online group.” We’re more online than usual these days due to pandemic social distancing, but more typically the partner- and group-oriented parts of our work are done face-to-face, or maybe over the phone.
As for your observations about Christianity, It could both be true that Christian virtues are not oriented to this world and our thriving within it, and that they nonetheless have the side effect of promoting worldly flourishing. Another possibility is that people who identify as Christians report better e.g. happiness for reasons that aren’t related to the practice of traditional Christian virtues (reasons like camaraderie, sense of purpose, reassurance about mortality, belief that the universe is benevolent, etc.) You suggest a potentially useful line of inquiry.
Beeminder is already a Slate Star Codex advertiser, obviously (the first one, I think! maybe tied with MealSquares?), but in case you’ve become banner-blind to it, let me reemphasize that BEEMINDER. All the cool rationalists are doing it! Act now, etc. Advertising!
It’s especially great if you’re struggling to stay productive while working from home. (As if it weren’t bad enough how we profit off your failure, now we’re profiting off of death and a global depression!) We have some recent blog posts about that and I generally include posts I think will be interesting to this crowd under the “rationality” tag.
I guess I should also clarify that I’m super kidding about failure and death. We have another elaborate blog post explaining why we in fact make money in proportion to how much awesomeness we induce in you (as would have to be the case for us to have been in business for, oh my goodness it’s been 11 years since I first wrote about the predecessor of Beeminder on LessWrong): ““Derailing Is Not Failing; or, Beeminder Revenue Proportional To User Awesomeness“.
In conclusion… please beemind things? If you like the idea but don’t know what/how, I can answer questions here!
Best app on my phone.
Well, best app on my Android. My new iphone is better in every way except beeminder’s app quality π
But it’s still valuable enough to use by miles.
Woo-hoo! Thanks so much for saying so!
As for the smartphone apps, I don’t know how soon we’ll close that gap but we are putting in more hours on the iOS app than the Android app at the moment so… maybe eventually your iPhone will fully Pareto-dominate your old Android! π
(I announced part of this a few weeks ago, but it was in a hidden open thread, and there is more news.)
My brother-in-law Bill died unexpectedly last fall. In the 1990s, he and his cousin Cecil published two science-fiction novels about Evan Larkspur, a poet and playwright who finds, after a century in space and cold-sleep, that his plays have become both revered and seditious. The circumstances of his return are such that his memories are jumbled, and he tries to make sense of the oppressive world to which he has returned, while sometimes fearing that he might be just another madman who thinks he is Larkspur.
I always hoped he would release these as eBooks; alas, he somehow never got around to it. I have made it a project to preserve his legacy by making that happen, and so have spent some time reconciling and editing the various minor revisions we have found in his files.
Both of them are now available on Kindle. The first, The Unwound Way, is here, and the second, The End of Fame, is here. (These are the American landing pages; both are available in English on lots of Amazon sites.)
I still think both are cracking great yarns, lively and funny, that say interesting things about identity and freedom without ever interrupting the action. Catherine Asaro gave End of Fame a lovely review at the time, excerpted as an Amazon review of the original paperback, and last week I confirmed that she still remembers the review and loved the book. If you are looking for a way to while away the sheltering-at-home hours, give them a look.
(At the risk of channeling Christopher Tolkien, I am currently working on preparing Bill’s first, unpublished novel, Tilt, for release on Kindle. It is unrelated to the two Larkspur novels, but to my mind at least as good. Watch this space.)
Thanks for making sure these got released. I’ll check them out.
I forgot to throw in a tidbit about End of Fame: it mostly takes place on a breakaway planet the hero visits, whose philosophy is “the strange idea that properly constituted government shouldn’t need hundreds or thousands of new laws every year — that after eight hundred years, it should rarely need new laws at all.” Their motto is:
I expect this will intrigue some of the SSC community and repel others, but you might as well know.
Heh. Reminds me of one of the more amusing scenes from Piper’s short story “A Slave is a Slave” (available, like most of Piper’s work, on Gutenberg), where the people from the Empire are asking the locals on Aditya about their government and how it works:
Not that the Adityan government, as we see later in the story, is an example to be emulated in other respects (the title of the story is a big hint in that regard…)
Hah! I can well believe that Bill got the idea from Piper, but I’ve always liked the (pat?) pithiness Bill’s formulation.
I wrote a forecasting newsletter for April 2020, and intend to have it go on for at least 5 more iterations. You can find it here, and sign up here
Thanks for signal-boosting https://coronainformationmarkets.com !
Apologies for a poorly-specified classified! I’m looking to (hire? contract with? just talk to?) someone who
– Has some knowledge of electrical engineering; and
– Speaks either Mandarin or Cantonese
My father-in-law is an electrical engineer who runs a small business in China manufacturing lighting control components of his own patented design. He speaks no English, but is convinced that his technology represents an advance over current practice for energy/cost efficiency reasons and that most municipal and industrial lighting projects in the U.S. would want to use it if they knew about it.
I’m a monolingual non-engineer and have no way to evaluate if he’s right about that. Probably every small business owner thinks something similar? But I previously asked for advice here a couple years ago, and some helpful SSC readers (thanks Alex!) took a look and said that it’s at least facially plausible. And that the U.S. could indeed be a better market for an energy-efficiency technology than China, where energy tends to be subsidized and efficiency is not valued.
So I’d love to find someone who could:
– Talk with him and understand the technology and its business case;
– Investigate how municipal (e.g. streetlights) and industrial/commercial (e.g. floodlighting) lighting projects get done in the U.S., as a business; and ultimately
– Recommend whether and how his technology could be marketed here.
This could be scoped as a consulting assignment with the information as a deliverable; we can figure out a market rate and I’ll pay you. Or if it does look promising and you were up for it, you could very likely sign on to help market or distribute the products for a salary and/or equity. Who knows, it could be a big hit?
On a personal level – he’s a reasonable, mild-mannered man and I suspect would be easy to work with. His daughter and I are hoping he will agree to move here to live with us and his grandkids, and finding a way for him to do business here would go a long way towards that. I’d be grateful to anyone who could help or could help me find help.
My email is lipseylc at gmail. Thanks in advance!
Hey, I think I we briefly talked before but I never followed up. I tick all the boxes except for the language bit, right down to municipal government contacts. (And I do have experience with Chinese people on a fairly convivial level, one of my past serious relationships was with a girl from Guangzhou, but I don’t speak the language well.) Still interested in helping, if that’s not an insurmountable barrier. Just reply you’re interested here and I’ll email you, from my personal email this time so it won’t get lost.
Absolutely I am interested! Knowing something about the market would be very helpful – thanks!
Sent.
[I may be able to serve as the third point of this budding triangle. I speak and read Mandarin as a reasonably-strong second language, and have experience translating news articles and literature from Chinese to English. I’m not an electrical engineer and have no experience with municipal governments, but I have software development experience and good technical-writing skills.
Perhaps @Erusian and I could band together to communicate with your father-in-law: for example, the latter might provide diagrams or writing which I translate into English for @Erusian to make comments upon before translating those comments back into Mandarin.
If you want to get in touch, email contact.canyon.fern@gmail.com, and I’ll return mail from my regular account.
-Ludovico (Canyon Fern’s human assistant)]
Thanks! I will email you at the weekend.
My friend from an SSC Meetup started a podcast type show. I think it’s gonna be about X-Risk and climate change and stuff.
He might be willing to interview people from here too! π
Since we’re all now spending our lives in video calls, launched an effort to fix everyone’s (crappy) webcam appearance: https://www.spruce.cam
Pairs you with an (underworked) professional photographer to spruce up your online appearance, either via a quick evaluation of your current setup, or a live consultation to find your best location.
Great idea.
Nice idea. Your website is slick too. You could try targeting advertising for people who Liked dating services like Quarantine Together or OKZoomer on Facebook.
This is awesome. I hope the first thing they say is ” in Zoom, go into settings – Video settings, and turn off ‘Touch up my appearance'”. I have seen some truly awful looking zooms.
The next thing is advice on Wi-Fi. Horrible latency and stuttering have I seen.
The advice boils down to “use a wired connection if at all possible. And it *is* possible, yes, really, it is. If you absolutely can’t use an ethernet cable, make sure *every single* wi-fi connected device in your house is in the same room as your access point (“router”). Not just your laptop, *everything*. Or use Wi-Fi 6, if that is impossible too.”
Used thanks to seeing it here, strongly recommend. I intend to fix the items noted, and then get a second round of advice.
I think your link text is broken; try https://twitter.com/willinthethrill
I am looking for feedback on my recent paper, ”Self-explaining AI as an alternative to interpretable AI“. The paper is about how the functioning of deep neural nets makes them fundamentally hard or even impossible to interpret in a human-digestible way, drawing on the recent discovery of double descent. The concept of “Self-explaining AI” is explored as an alternative to traditional methods of interpretation. I can be reached at daniel.elton@nih.gov.
As a follow-up on this work I have been thinking about and researching how to ensure robustness to distributional shift, as well as adversarial examples. It would be nice if when an AI is asked to work on data which is outside its training data distribution, a “warning light” activates to warn the user the AI may not function (since deep learning can’t extrapolate). It seems this has been studied before, using anomaly detection methods to monitor the input data. I think something like this is needed when using AI to detect COVID19 on Chest X-Ray. There are many papers on deep learning models trained to detect COVID19, but it’s known the transferability of such models across different datasets (ie. different scanners, image resolution, etc) is very low. This hampers the real world utility of such models greatly.
Whenever I hear about training AI which can explain its decisions I fear that it will push it towards solutions similar to what one can see in human brain consciousness, public relations department, or courtroom, where explanations are biased toward being accepted by the audience to the point that they might be disconnected from the truth (some suggest that people are unconscious of thy true motives so they can more convincingly talk about made up motives). In the context of alignment of an above-human GAI I am worried that a super-intelligent lawyer could persuade me to whatever they need.
How is that addressed in research?
Perhaps such an AI is more of an persuader AI than a problem solver AI.
I make melodic rap, in French.
I already have more than enough advice and feedback; but it you just want to listen, enjoy:
https://www.youtube.com/watch?v=E4ps2E-2bFY
Feel free to subscribe too!
I listened, and I enjoyed it very much!
I’d never listened to melodic rap before. I sort of respect rap, but don’t really like it. Now I know why. Playing around with your voice, adding subtle modulations, etc; those are my favorite aspects of singing and songs. I really like how this comes together with the minimalist percussion to create an airy, mysterious feeling.
This is cool. I like the cover art a lot, too.
And your including the lyrics in the video is helping me practice my French listening. Very cool!
It reminded me of a British rapper (or singer?), yxngxr1. If you wish to check him out, the songs I like best of his are Riley Reid and Iann Dior.
I write a blog called Unexpected Values. Check it out here!
Some highlights:
– How to test and trace if you don’t have enough tests (About pooled Covid-19 testing)
– Beyond the mean, median, and mode (My most popular post; somewhat math-heavy)
– Finding Shawn Mendes (Hopefully humorous)
– My favorite puzzles from the 2020 MIT Mystery Hunt (Might be interesting if you’re into puzzles)
– Lofty Expectations (Pretty speculative; I argue that expected values for everyday things are infinite, and here I talk about the implications of this for decision-making)
– Predictions for 2020 (In the style of SSC, but made in December)
I also just published my own puzzle, and am offering (small) cash prizes for the first people who solve it!
This made my day, especially the puzzle stuff.
Thanks, glad to hear that π
I found this perspectives on means, medians, modes, and their generalizations to be extremely fascinating and very readable! Have you thought at all about their applications to decision theory or social choice theory? (Given a utility function and a probability distribution), the expected utility of an action is just a weighted mean, after all, and choosing the outcome with highest median utility is also a fairly robust decision-making process (which provides the inspiration behind social choice techniques like majority judgmentβand, in a certain framing, approval voting). Could this perspective be used to construct a family of βtypical utilityβ measures that interpolate highest median and expected utility, and extend past both?
Anyway, I loved that article and will probably be back for more.
Thanks for the kind words!
I’ve mostly thought about this in the context of aggregating opinions rather than decision theory or social choice, but that’s an interesting thought! This is interesting from the perspective of things like Pascal’s mugging, because your expected utility is higher if you give into the mugging (I would argue), but your median utility is higher if you don’t give in. Since many people’s intuition is that you shouldn’t give in, you might ask whether there’s some sort of interpolation between the mean and median that reflects how people feel intuitively about decision making.
As for social choice, that’s also interesting. Let’s say n people have to make a decision between one of two choices, and each person says what the signed difference in utility between the two choices would be for them (i.e. their utility under choice 1 minus their utility under choice 2). Now we have n numbers that we want to aggregate. There is a clear reason to take the mean (and then go with option 1 if it’s positive and option 2 if it’s negative): that’s what you do to maximize the total utility of the n people. But note that this is only the case if everyone tells the truth. On the other hand, if I slightly prefer option 1 to option 2, I could say that I like it way more just to make sure option 1 gets chosen. So this isn’t incentive-compatible. On the other hand, take the median of the n numbers is incentive-compatible, so that’s a reason to take the median! I wonder whether in practice going with an interpolation between the two, as you suggest, might be reasonable in some settings.
I don’t agree that maximizing median utility is a robust decision-making process. It regularly gives bad answers to common decision problems involving potential large risks or large rewards. For instance, you can slightly increase your median utility by not using a seat-belt next time you drive.
I am very very close to solving the puzzle but I can’t quite figure out how to translate my results into a word or phrase. If anyone wants to collaborate please email metacelsus@protonmail.com
MD5 hash of my partial progress, just in case I need to prove it later: 8f33cb1fa7291225f7b9becc13a5ee91
This is great! The post on p-medians in particular is fascinating.
I bounced on Lofty Expectations, though.
An important claim therein is βGrahamβs number is rare among large integers because it is computationally simple.β If we are saying Grahamβs number is rare, this is just to claim that we can actually know that other large integers are not computationally simple.
But this reasoning requires that P = NP.
It is easy to verify that Grahamβs number is computationally simple – but it would have been much more difficult to start from Grahamβs number and prove that it was computationally simple (IE to use it to discover Knuth notation).
It is the same for other large integers. It would be hard to prove that any large integer is computationally simple, even if it would be easy to verify that computational simplicity. In other words, there may be operations that are as simple and generative as Knuth notation, but that remain undiscovered, and we do not know which operations would reconstitute these large integers.
The only way we can claim, a priori, that certain large integers are computationally complex is to presume that there are no simple operations that would generate them and that we should be able to discern this just by looking at them. IE, that it would be just as easy to determine that complexity as it would be to verify that complexity.
I agree with everything you wrote in the last three paragraphs. But I don’t think this points to a flaw in my argument. I agree that for any given number, it’s impossible to prove that its Kolmogorov complexity is large. However, I can prove to you that most numbers of the same order of magnitude as Graham’s number have a large Kolmogorov complexity. Observe that there are 2^k Turing machines of length k. This means that there are at most 2^k numbers with Kolmogorov complexity k. This guarantees that among the numbers {1, 2, …, G} (where G is Graham’s number), at most 2^(k + 1) have Kolmogorov complexity at most k. Now set k = log(G) – 2 and you’ll find that at least half of the numbers in {1, 2, …, G} have complexity greater than log(G) – 2 (which is an enormous number).
So yeah, for a given number it’s impossible to prove that the probability that there are that many people in your room is astronomically small. But we know that for most of the numbers the size of Graham’s number this is the case (and, I argue, for Graham’s number it’s not the case).
This is fascinating. Iβm of two minds.
β-
I buy the Turing machine logic. Especially the k = log(G) – 2 calculation, which is very elegant. And there would be a ready connection to the physical world : if the world is basically made of Turing machines, and if a computationally-simple Turing machine is less costly than a more complex one, it would make sense that these simple machines (these simple numbers) would be physically probable.
But there are many such numbers, taking the Knuthβs up-arrow notation form of x[^y]z. I donβt know if this group of numbers has a name, but why not call them Knuth numbers? 2[^1]3 = 8 is a Knuth number. 5[^2]2 = 2.98E17 is a Knuth number. If computational simplicity is an important driver of a numberβs prevalence, then Knuth numbers would give some probabilistic structure to numbers.
Moreover, there would be a great many Knuth numbers simpler than G. The simplest Knuth number is 1 – especially because 1[^y]z = 1 for all values of y and z. So maybe I should conclude that there is indeed only me in the room.
β-
However, I do remain skeptical that we can use this kind of computationalism to connect numerical structures to physical probabilities. This is starkly counterintuitive. Intuitively, I wonder if we are conflating the computational complexity of explanations/stories with unrelated mathematical structures. (βSo what about the stock market?β)
Also, how do we apply this reasoning to irrationals? CMWIW here. The computational complexity of calculating pi to the nth digit is O(M(n) log n). As n approaches infinity, the computational complexity approaches infinity. If computational complexity determines prevalence, then we ought not be able to observe pi – or it should appear to be rational.
Right, so this is an argument that your prior that there’s 1 person in your room should be pretty high. But similarly I argue that your prior that there’s G people in your room should be kind of high as well (much lower than that there’s 1 person, but on the order of magnitude of 10^-1000 rather than 1/G).
Say more? I’m not sure what you mean.
I don’t think that the time it takes to solve a problem has a particularly close connection to how easily you can encode the problem (i.e. write a Turing machine that solves it). For instance, one could pretty easily write down a Turing machine that computes the G’th digit of pi, but the Turing machine would take a huge number of steps to finish.
If we consider all computationally-simple numbers to be of equal probability, your estimate would entail that there were only 10^1000 computationally-simple numbers in the set {1, 2, …, G}. G is far too large for that to be true.
Moreover, many of those numbers will have a smaller k than G, meaning the Solomonoff estimate of their prior will be greater than that of G. Normalizing over all simple numbers, P(G) would be dwarfed by the probabilities of far more numerous more simple numbers.
So, intuitively, I could buy that P(G) > 1/G, but not by much.
But there is some kind of connection here. We can easily encounter infinite digits of pi, even if it would be impossible to calculate infinite digits of pi (and that even if the algorithm is simple).
So there must be more to it than Kolmogorov complexity of our best descriptions.
You have got to see Aronofsky’s Pi
There is a simple Turing machine that outputs Graham’s number — certainly one with fewer than 1,000 states, and there are on the order of 9000^3000 such Turing machines. So if you’d like, change 10^1000 to 10^15000 or something; the point still holds.
Right but the way to normalize is to divide by 2 to the power of the complexity, and 2^(-10^1000) is way larger than 1/G.
What do you mean?
Well, alright then. Solvitur ambulando.
I really enjoyed “Beyond the mean, median, and mode”, thanks for sharing it!
I don’t really like the motivating example in ‘Lofty Expectations’.
Let’s say you hand me a solid ball and ask if I think it can turned into two solid balls of the same size. I say ‘No’, and you produce a proof of the Banach-Tarski paradox. I’m not going to be satisfied, because you’ve started talking about strange abstractions like ‘sets’ and ‘the real line’, whereas the the original question was referring to a ball I was holding, which is a very different sort of thing entirely.
Similarly, when you ask ‘What is the expected value of the number of people in the room’, it’s taking for granted my understanding of words like ‘people’ and ‘room’, and my understanding of the words ‘people’ and ‘room’ precludes there being graham’s number of people in a room. It seems strange to have to consider the probability that I’m fundamentally misunderstanding the concepts being discussed, because that seems like it could dissolve any part of the question. Perhaps I misheard and you said ‘Variance’, it’s not impossible!
I think it’s better to present the Banach-Tarski paradox as a surprising result in measure theory rather than a Weird Trick For Doubling Your Bars of Gold, and I think this blog post would be better presented as a neat result about a nice distribution of integer-valued random variables rather than a suggestion that we’re currently thinking about Expected Value in a fundamentally misguided way.
(You address this objection in a footnote when you say that ‘you can take any definitions you want β so long as it isnβt mathematically impossible for there to be Grahamβs number people in your room’. But I think that’s like saying I can take any definition I like of a solid ball so long as you can apply an action of a free group on two generators to it)
Yeah, this is fair. I don’t really have a good counterargument to what you said. Do you think this is just a property of the specific motivating example I chose to give, or do you think that literally any real-world example would have this property?
For instance, what if I said something like “the number of atoms in the universe, conditioned on it being finite”? I would argue that it’s pretty reasonable and intuitive to think that the expected value of this quantity is infinite. But at the same time you could argue that the possibility that there are Graham’s number of atoms in the universe is at odds with your understanding of what “universe” means. There may be no clean argument like “people are about 100 liters large and my room is about 100 thousand liters large so there can’t reasonably be more than 1000 people in my room” but if you were a physicist you could probably come up with a similar sort of argument. And that argument would be right if they were 100% confident about the structure of the universe, and they could argue that if that’s not the structure of the universe then you mean a totally different thing by “universe” than they do, but somehow that feels wrong to me. Is there a fundamental different between the “number of atoms in the universe” thing and the “number of people in your room” thing? Maybe, but it’s not clear to me that there is.
(But at this point I feel like I’m doing the sort of sketchy/pointless philosophy that I’m not a fan of, so maybe the previous paragraph is totally silly because maybe it’s just an argument about the definition of “definition” or something like that.)
I think there’s an impossible-to-bridge gap between the real world and abstract reasoning that tends to doom these sorts of things…
Questions about the real world like ‘What is the expected value of the number of people in the room’ are invariably wooly, and depend on tacit agreement between the involved parties on the underlying meaning. When we agree to answer a question with a piece of mathematical reasoning, the key part is agreement on how to phrase the question as a statement about mathematical objects.
Sometimes there’s a vague sense that the re-formulation of the question as a statement about mathematical objects is actually fine, that you haven’t lost much in the translation. Say you’re at a late stage of a chess game, and you want to prove that your opponent can’t possibly win. You might regard the chess board as a set of states that you can move between according to certain rules, and might convince your opponent that certain states cannot be reached from the given starting position. We haven’t proven anything about the physical pieces on the chessboard in front of us, but playing chess is more about moving between states according to certain rules than it is about physical pieces on a chess board, so most people are happy to accept this argument.
At the other extreme is something like the Banach Tarski paradox. Even if your correspondent is initially willing to accept your suggestion that you represent a physical ball as a certain subset of R^3, they’re going to object once you start exploiting that formulation to start doing things that you really can’t do with a physical ball.
I think your example is somewhere between the two! There’s definitely a nice agreed-upon convection for rewriting the words “What’s the expected value of…” in terms of mathematical objects. But the part where ‘people’ and ‘room’ get written in math-y terms seems to smuggle in some trickery where we represent them in ways very distinct from how we actually think about them (such as there potentially being graham’s number of people in a room).
And I don’t really think swapping in atoms improves it that much. People certainly have much fuzzier conceptions of what atoms are compared to people and rooms. But if anything that makes it more difficult to translate questions about the quantity of atoms into mathematical statements in such a way that people are happy that the resulting mathematical formulation actually captures the meaning of the original question.
I run the twice-monthly Mozilla Hubs meetups you may have seen linked lately.
I have been trying to get an speaker for each meetup to do a 20 minute talk with 10 minutes for questions. I have the next three booked (Scott Aaronson will be our featured speaker this weekend) but am looking for more for this summer. If you have any suggestions for speakers who you think would be both interested and interesting, or would like do do this yourself, please email me at ghenlezo1@gmail.com.
I wrote a superhero web serial called ‘The Fifth Defiance’. I’ve advertised it in these threads a time or two, thought I’d give it one final go. Thanks to all who read!
https://thefifthdefiance.com/2015/11/02/introduction/
I picked it up after seeing it advertised in a past thread, and I’ve been enjoying it ever since. Thanks!
Same here! Put it down for a while after I caught up to the in-progress point but I liked it a lot and I’m sure I’ll get back to it sooner or later.
Do you have the ability to throw it on goodreads? I’d love to add this to my “to read” list so I don’t forget about it.
I am adding this to my to-read list, or at least my to-try-reading list.
It’s pretty good so far (I’ve only caught up to about 2018 or so), but is there some way I can read it in a more conventional form ? The blog format includes lots of interruptions in the form of “sorry for the late post” or “please vote for us”, etc. They make perfect sense given the format, but still, I find them pretty distracting.
I have recently launched a community for people interested in engaging in dialogue about leadership, strategy, critical thinking, probablistic decision-making, etc. My own focus is on strategy consulting and executive coaching for high-growth tech startups. But all interests and businesses are welcome. It isn’t a sales platform for my services or anyone’s. It’s for forming relationships and engaging with like-ish minds. I’m also going to be hosting some Mastermind groups and virtual co-working through the platform — and members will be given the option to participate. Virtual co-working opportunities will be free for the foreseeable future.
If you’re interested simply go to https://BeyondBetter.Online and request to join.
If you want to know more about who I am go to https://BeyondBetter.io
Firefox on macOS is giving me an error about how it failed to make a secure connection.
Same. Firefox on Windows.
Same, Firefox on Arch Linux.
Same for Firefox on Android. I wonder what SSC’s favorite browser is?
Same, Firefox on PDP-11.
Bet getting Firefox to run in 64K split I&D space was an adventure.
We were tired of broken donation flows, getting boatloads of email spam, and spending hours scrounging up donation receipts at the end of the year, so we built a better way to donate to the nonprofits you care about at https://www.every.org. You might have seen us mentioned in the April edition of the EA newsletter.
Some links that people here might find interesting:
COVID related causes
GiveDirectly
The Against Malaria Foundation
Our FAQ
Could you say why it’s better?
I’m not them, and had the same question. The top of their FAQ gives what’s presumably the “official” answer, and it would have lowered friction for me if it were quoted here rather than linked, so I’ll copy-paste:
* It’s easy β Manage all your giving in one place, get one tax receipt at the end of the year, and never forget about a recurring donation. In fact, you can pause and cancel those with one click.
* It’s privacy-first β You donβt have to share your contact information with a nonprofit in order to support them.
* It’s secure β Your donation transactions are secured through Stripe and Plaid, industry leaders in online payments processing. No need to worry about unstandardized payments.
You can maximize your impact β If you donate with your bank account, 100% of processing fees are covered.
* It helps nonprofits reach more people β People are more likely to support nonprofits that other people support. When you give on Every.org and share your activity with your community, you’re effectively supporting your nonprofit twice with the ripple effect of giving.
* Itβs privacy-first β You donβt have to share your contact information with a nonprofit in order to support them.
OH MY GOD this shouldn’t be such a problem but it’s a huge one and if they solved it that’s wonderful.
Donating to a charity gets your name put on a sucker’s list that gets sent out to every Tom Dick and Harry in the world. I had an idea to start a study to figure out which charities are the worst and shame them. But it looks like these guys have found a different way to solve the problem right now.
Probably the biggest psychological barrier to me donating to any charity is “do I want to subscribe to a lifetime of shit in my mailbox” and they’ve fixed it. (In theory. I haven’t used them yet but I’m just ecstatic that someone has just announced they’ll clean up the pile of vomit that’s been on the sidewalk for years.)
Donating to a charity gets your name put on a suckerβs list that gets sent out to every Tom Dick and Harry in the world.
Seconded. Even worse is the “donate once, be chump enough to give contact details, prepare to be able to wallpaper your entire house for the rest of your life with begging emails and junk mail from that organisation”.
I’m generally hard-hearted enough to stick the emails onto the junk filter so they’re automatically dumped and the irritation factor isn’t quite enough to overcome the sense of duty towards charitable causes, but it is annoying (surely to God the amount of money wasted on mail shots that go straight into the recipients’ bins would be put to better use for the causes that they are begging donations for?)
This is why you give a small donation once to charities you hate.
Thank you for copying this over, Elena! Thought I set it to send me email notifications about responses to my post, but I guess I messed that up :/
One additional point to add that I don’t think is super well reflected in that answer in our FAQ is that in the interest of increasing the efficiency of your donation, we don’t charge any platform fees. This means that when we say there are no processing fees when you give with a bank account, truly 100% of your donation is going to the nonprofit. Payment processing fees for most credit cards are 2.2% + 30c (stripe) + 2.25% (NetworkForGood) for a total of 4.45% + 30c, and we’re continuing to explore different ways of bringing that down.
For those wondering how we can charge no platform fees – we’re a nonprofit with ops funded by our own donors for now. In the future we might consider doing something like GoFundMe where we ask for an optional “tip” on top of donations, but we are committed to NEVER taking a cut of the donations themselves.
I already use a DAF, which seems to cover a lot of the same points. Is there any advantage I’m missing over a DAF? It seems like the main one is that Every seems to work out better at lower amounts (I pay a fee of a few bps per year, but the minimum makes this significant at lower amounts, and there are minimum donations and the like).
This looks very interesting; I have a few questions.
From a technical back-end perspective, this looks like Every is running its own DAF; is that right? In particular, that was my inference based on the portion of the FAQ talking about how if you donate to charity X, and the donation turns out to be undeliverable (e.g. because charity X has ceased to exist), that Every will provide equal value donation credits (rather than a refund). That would make sense to me if Every was a DAF, so I was curious.
Second, on a related point, does Every support the ability to make deductible donations that are subsequently allocated out to individual charities? Put another way, would Every permit people to make the type of multi-year bunching donations that DAFs are sometimes relied on as a tax optimization strategy? E.g. donate $24,000 today, and then donate $1,000/month to individual charities for the next two years. That would definitely be of interest to me if so — I get that that doesn’t seem to be your primary marketing thrust, but I’m curious if that’s something Every’s platform would permit.
Third, less of a question and more of a comment, your credit card processing rates are equal with the best I have found (I’ve spent a decent amount of time looking into credit card processing fees for nonprofits). From a pure optimization standpoint, I think there’s more value in facilitating donations via credit card than people tend to think for a very important reason: the entire donation (including the processing fee!) is deductible. Thus, if you have someone itemizing at a 30% marginal tax rate, the 2.2% fee becomes a 1.54% fee, and getting a credit card that has better than a 1.54% reward rate on general spending is not difficult. This effect becomes even more pronounced at higher marginal tax rates (e.g. it becomes a 1.32% effective fee at a 40% marginal tax rate).
I’d never donated to anything before I tried Every a few months ago. Seemed pretty cool.
Are you tired of having to training your multi layer neural nets in high precision floating point? Do the gradients of your 1000 layer convnet keep vanishing? Want to stop paying NVIDIA-Mellanox so much money for floatingpoint matmul hardware and networking? Do you like turning high precision, centralized problems into low precision, embarrassingly distributed problems?
If so, lets talk!
You can email me at isaac@isaacleonard.com
I don’t have good answers to any of these problems, but I’d love to talk about them with you.
Now I can’t help but imagine this as a therapy session:
“When you’ve found out that the gradients have vanished, how did it make you feel?” etc.
I wonder if something like that could actually be in demand, especially a therapy for webdevs.
They don’t make pills for framework fatigue, but I want some.
Have you looked into Google’s approaches to these problems? Their TPUs generally use low precision.
I’m familiar with Googles TPUs. bfloat16 helps, but makes vanishing gradients even worse. They use other hacks to work around this, but it’s not a complete solution.
16 bits per weight/activation is excessive for inference. If we are clever, we can quantize already trained models to binary weights and activations with relatively small loss to accuracy. https://arxiv.org/abs/1603.05279 Why can we not train directly in binary? (I know perfectly well why we cannot but please explain it to me like I do not.)
The whole field of end to end gradient descent has fundamental scaling limitations. It does not distribute well. (I know that people do distribute end to end gradient decent training, but they need a big pile of hacks and it still does not scale well.) By distribute well, I mean efficiently train very large models on 100,000 cheap spot instances across dozens of data centers on different continents. Google’s TPU push against the problem, but they do not circumvent the fundamental limitations.
We can train some models one layer at a time: https://arxiv.org/abs/1812.11446 I have not seen much interest in this since it was published. Efficient-market hypothesis says that this means that greedy layer wise training is not actually very useful.
A TPUv3 pod has BRUTE STRENGTH, but it is already beginning to push against theoretical limitations of cooling and data transfer speed. Unfortunately, in our world, BRUTE STRENGTH cannot violate thermodynamics.
Forgive me if this is a stupid question, but why can’t one prevent the gradients from vanishing by normalizing the gradient vector every few layers when doing back-propagation? (So instead of passing back the actual gradient vector to the previous layer, you pass back a vector that points in the same direction, but has a length of 1.)
I’m not very familiar with gradient normalization, but that sounds reasonable. Does any one else here have experience in this area?
This feels ugly however. We are adding hacks to fix problems caused by other hacks. I wish we could do away with the while mess and do something that feels more elegant.
A thing feeling ugly does not not necessarily mean that there exists a more elegant alternative, but sometimes it does.
There are several different approaches to this issue. One is to use activation functions that don’t have these problems, like ReLu. Another is to normalize (although Phi’s suggested normalization is probably a bit too aggressive, hence something like batch-normalization, which conserves some information about the relative lengths of gradients). Another is to construct gradients that are additive, such as in LSTM models. I don’t think it is an enormous problem for the field.
Sorry to come late to the party. This is by no means an authoritative answer, but my intuitive understanding is that this would prevent generalization by preventing the lower layers from stabilizing. Once you have trained to a certain point, you need there to be smaller and smaller updates settling down into the lower layers. It’s the same reason competitive AI practitioners use differential learning rates when fine tuning through transfer learning… you start by freezing the whole convnet backbone, training only the new head, then when you unfreeze you set exponentially lower learning rates for the lower layers than for the upper… you need to retain the stability of the lower layer kernels.
“16 bits per weight/activation is excessive for inference. If we are clever….”
I think insistence on being needlessly “clever” is a notorious source of software bugs. Often software should be as simple as it can get away with, because often “cleverness” makes it harder to understand, and results in unrecognized inefficiencies instead of the moderate recognized inefficiencies that the “clverness” was trying to avoid.
What you say is very true.
In this case, replacing floating point with binary activations and weights allow us to replace the equivalent of 64 floating point with an XOR, a popcount and an unsigned integer add. It also reduces data volume by a factor of 32 which is nice for memory bandwidth. Binarization of weights and activators is a _big_ performance improvement.
Converting an already trained floating point model to binary weights and activations does required quite a lot of cleverness of the bug inducing sort. If we naively quantize, there are big accuracy costs. There are various schemes to quantize as you train, or train in a hybrid of float and binary, etc, but they tend to be rather complex.
https://arxiv.org/abs/1602.02830
https://pjreddie.com/media/files/papers/xnor.pdf
https://arxiv.org/abs/1809.03368
https://www.mdpi.com/2079-9292/8/6/661/pdf
At inference time, binarization of weights and activations is essentially free performance. It is easy to implement, and offers big constant factor performance improvements. Why is it not universal to binarize models before inference when it is such a free lunch? I’m guessing a big part of the reason is that it makes training significantly more complex, and, as you point out, the cleverness needed to get it to work usually outweighs the performance benefits. And so, people run inference in float, or quantize to 8 bit which is a lot easier, and (apart from XNOR.ai) ignore the big performance benefits they could be getting.
Part of my desire to train directly in pure binary is to get the binarized models without the complexity of current training methods. I want to avoid the cleverness which is currently needed to create binary weighed models, and instead, find some boringly simple way to directly train multi layer NNs with binary weights and activations.
I don’t have answers either, but maybe this paper will inspire something: Random feedback weights support learning in deep neural networks
Thank you, this is an interesting paper which I had not encountered before.
I’m a bit disappointed that they didn’t test it on CIFAR10 conv nets or similar.
I recently started a Journal of Unfashionable Ideas modeled after the Journal of Controversial Ideas. I’m looking for fellow undergraduates who are interested in submitting work and/or editing it with me.
If you’re interested, email me at isadore.johnson@uconn.edu
Seems like a good idea. Is the idea that you submit under a pseudonym?
I applaud the fact that you started this. What’s your plan for dealing with:
1) The inevitable large number of highly racist papers that will be submitted, and,
2) The backlash that will result no matter what your decision is regarding (1)?
Also, to reduce your journal’s being misperceived as barbarically “conservative,” you should make an effort to find and present some unfashionable ideas that most readers will see as “liberal.”
Hello, we have just created a course on Udemy, “Communication for Remote Teams”.
It is 2 hours long, and currently on sale for $9.99.
https://www.udemy.com/course/communication-for-remote-teams/?referralCode=CC146E89E7FAD03E7A21
It is for anyone who is working on remote teams and want to communicate better, to increase team efficacy, accessibility, culture, and productivity.
Thanks for checking it out and forwarding the link on to anyone else who might be interested.
Learn more about Remotor Consulting Group: https://www.remotorconsulting.com/
As someone who is quitting a job after only 3 months due to finding myself on a global team with wildly dysfunctional communication, I think I’ll be checking this out. Honestly my goal is to get back to being on a co-located team, but I think some amount of remote work is soon going to be inevitable, so getting better at it or at least being able to sniff out a bad situation while still in the interview phase would be of major value to me.
Thanks!