Please Vote For ACC Winner

I’ve now posted all eight adversarial collaborations.

In case you missed any, you can find a list of them (with links) here.

If you have read all the collaborations, please vote on your favorite. This year I will decide the winner by popular vote; I don’t feel like putting my finger on the scale this time. I will give $2000 to the first place winner and $500 to second place. You can vote for your favorite collaboration here. No, you may not vote for the Grinch.

Thanks again to all participants, readers, and voters.

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

54 Responses to Please Vote For ACC Winner

  1. Steven says:

    Personally, I started by defining a scale from 0 to 10, to score them on:

    0 Actively harmful to society, muddies the discussion and presents bad faith arguments
    1 Unhelpful, muddies the discussion, and dwells on irrelevancies
    2 Very low effort – one side of the collaboration is asleep at the wheel
    3 Low effort – two arguments presented, but no attempt made to actually collaborate
    4 Acceptably written with major omissions/inaccuracies – topic not fully explored
    5 Well written, but some major omissions/inaccuracies or acceptably written with moderate omissions/inaccuracies
    6 Well written, with some moderate omissions/inaccuracies or acceptably written with only a few minor omissions/inaccuracies
    7 Clear, well written and reasonably sourced, with only a few minor omissions/inaccuracies
    8 Clear, well written and well sourced exposition of all important considerations
    9 8 plus thorough detailing of all areas of agreement, all remaining disagreements and any double cruxes found
    10 The platonic form of an adversarial collaboration

    Then I gave each one a score:

    Circumcision 1/10 I was tempted to give this 0/10, but based on my scale above, I didn’t want to assume bad faith. I would suggest that this is worse than the vaccination one last time, in terms of giving a smokescreen of legitimacy to a problematic idea. At least the comment by briandavidearp was a beautiful demonstration of Cunningham’s Law in action.

    Meat 6/10 Very nice write-up. Love the spreadsheet. Didn’t really address ethical meat sourcing (e.g. RSPCA certification in the UK)

    Calories 5/10 A bit too all over the place – seems like a difficult topic, due to the generally poor quality of research that is out there. No mention of things like intermittent fasting, which might convey similar benefits without causing prolonged weight loss.

    X-risk 8/10 Very solid – not much to complain about. I liked the fact that people’s starting positions were made explicit, and any changes of thought were detailed in the conclusion – gave it more character than a straightforward literature review

    Gene editing 6/10 Really interesting and informative. Did seem to have a few inaccuracies regarding the current state of medical progress, and seemed a bit too focused on medical interventions rather than giving space to the more general question around elective modification

    Abortion 3/10 Basically just a surprisingly civil internet argument, formatted like an essay – people talking past each other, and failing to identify any cruxes or areas of agreement

    Automation 7/10 Shorter than most, but well sourced. Might have benefitted from a bit more economics, as it was much more focused on the AI progress angle, but this presumably reflects the specialisms of the authors, so is not unreasonable

    Spirituality 6/10 A bit dry… I realise that the aim might be to put together something that looks like a literature review, but this goes a bit far – I feel like the question itself may not have been sufficiently divisive

    Based on this, I voted for the X-risk collaboration (though if multiple had ended up tying for first place, I would have voted multiple times, once for each, because Approval Voting is good!)

    I feel like a good way to do this in future might be to give people the ability to mark each out of 10 and just sum the scores, which would then be the same as Range Voting. Slightly more complicated than Approval Voting, but a very good system that simplifies to Approval Voting when people are maximally strategic.

    In general, I think it adds a lot to have people’s starting positions and ending positions detailed in the collaboration – it is nice from a transparency perspective, and makes it read slightly differently to just a plain literature review, which can be a bit sterile.

  2. RLM says:

    Scott, In the future, I would suggest you include required questions of the contestants so we can get a “behind the scenes” look at the process that produced the paper.

    Ask them things like “What were your beliefs beforehand?”. “Why did you enter this collaboration?”, “What’s your background?”, “How much time did you spend, and how did you two interact with each other to produce the paper?”

    I think that doing things like this will enable the readers to understand whether it was a legitimate adversarial collaboration, or just a one-person perspective (intentional or unintentional) under the false title of an adversarial collaboration.

    At the end the participants should also each write a letter of endorsement for the article saying that they feel they contributed to it as much as they felt was necessary and that it accurately represents their thoughts. This will help to determine whether they participants really put in the time or whether one ended up ghosting the other.

    Finally, special attention should be paid to whether any of the two participants changed their minds. If so, why did they change their minds? If not, why did they not change their minds?

    It’s good to have the articles present a unified front, but it’s also good to have a place for each author to speak in their own voice. Right now it’s hard to determine whether :

    – the two participants did through research, had commensurate backgrounds, and had a variety of spirited and useful debates about the issues
    – one person tried hard and the other ghosted
    – one person steamrolled the other
    – both people were really on the same side from the beginning
    – any changes in the participants’ beliefs happened
    – one person ran out of time and couldn’t put in the effort that they wanted

    And I think that knowing about ALL of the above is quite important. In addition to the benefits to the integrity of the ACs themselves, getting this sort of info would help to improve future ACs, since the future participants could use the debate style and setup of the most successful ACs for their own ACs.

  3. Aapje says:

    I was rather underwhelmed by the entries. None of them feel worthy of the win, IMO.

  4. DinoNerd says:

    *sigh* At least one of the entries pushed my buttons hard enough that I feel incapable of evaluating it fairly. I’m not going to vote in this one. I am going to attempt to read all of them, or at least skim them. My thanks to everyone for the effort involved.

  5. Quixote says:

    What’s the deadline? Ie how much time do we have to vote? I was traveling during the holidays and am a little behind in my reading because of it.

  6. hnau says:

    So I did reviews / subjective ratings of all 8. Collecting links (mostly for my own benefit):
    5/10 for circumcision
    8/10 for meat eating
    7/10 for calorie restriction
    4/10 for space colonies
    3/10 for gene editing
    6/10 for abortion
    7/10 for automation
    7/10 for spiritual experiences
    Looking back on these I don’t feel the need to update any of my ratings (though I was slightly shocked by how short the circumcision write-up is). They’re low-confidence subjective estimates comparing apples and oranges, and I feel like I could easily go plus or minus 1 on any of them. I still consider the meat-eating collaboration my favorite and, given the chance, would probably pick the spiritual experiences collaboration for second place (yes, the write-up is eye-glazing but that team seems to have gotten the most value out of the adversarial-collaboration setup).

    Many thanks to all the collaborators, and to Scott for putting this together! I learned a lot from it.

  7. JohnBuridan says:

    I really enjoyed this year’s contest. I thought all entries were worthy of being posted on SSC, and discussion in the comments sections were for the most part quite good. The SSC commentariat is an exacting group of commentators.

    Thank you all writers for being the grist to this unique community. I appreciate your resilience. 🙂

  8. sty_silver says:

    This is probably a coincidence, but I found the first half of the collaborations (definitely 1-4, maybe 5) quite strong, and the remaining ones much weaker. The abortion one was purely deontology so I’m not interested, and it also seemed to violate the basic rules of ACC (I don’t think that means it shouldn’t have been included). The automation one failed to talk about any of what I consider the strong arguments to be worried, which is a shame because it’s the most interesting topic to me. And the final one just seemed to be taking past me, even though I’ve been to a meditation retreat before. I don’t think there actually was a question on debate that I’d consider non-obvious.

    I ended up voting for the calorie restriction one, but the meat-eating one was also excellent.

  9. JN340 says:

    In case anyone’s interested: the New England Journal of Medicine (widely considered the top journal in medical research) recently posted a review article on intermittent fasting – strongly in favour. You can find it here, although it’s behind a paywall:

    Key quotes:
    “Many studies have indicated that several of the benefits of intermittent fasting are dissociated from its effects on weight loss […] these include improvements in glucose regulation, blood pressure, heart rate, the efficacy of endurance training, and abdominal fat loss”
    “Repeated exposure to fasting periods results in lasting adaptive responses that confer resistance to subsequent challenges”
    “Intermittent fasting seems to confer health benefits to a greater extent than can be attributed to just a reduction in caloric intake”

    Anyway, it’s got me doing IF again. (And yes, I’m voting for the IF AC.)

  10. Mark V Anderson says:

    I haven’t voted yet, but right now I am planning to vote for the meat eating one. I really don’t care how well they matched Scott’s adversarial rules. To me what counted was how much did I learn from them. I was very impressed by them in this. I learned something new from each of the first six ones; only the automation and the spirituality ones I learned nothing new (and even those were interesting). I am planning to vote for the meat eating one because I learned the most.

    I was originally going to vote for the abortion one because it was most likely to change my mind; but then when I tried to follow up with two of the footnotes, neither of them led to where they should have. I think having good footnotes is a requirement of a good research project.

    • sty_silver says:

      Have you checked footnotes on the ACs you liked?

      • Mark V Anderson says:

        No. I probably should do that.

        Edit: I went back to the Meat eating one. They have links, not footnotes, the the ones I was interested in, namely about fish not being conscious and cows in the pasture, both worked.

  11. pjs says:

    Something about the phrase ‘adversarial collaboration’ has – until now – biased me a certain way, and so I’ve been unreasonably disappointed in last years’ and this. But I’ve changed my mind.

    Think of these as a literature review concerning controversial (and hopefully, though not always in fact) _interesting_ questions. Who would you rather have produce such a review? Just one person? Two random people? What about two reasonable people chosen to have notably distinct biases, so each can and will call the other out for bad reasoning, but are otherwise sensible and collegial? It seems clear to me that 88% of the time, the last is going to get you the best review. Even though it might not look very fighty or adversarial, nor reach a particular conclusion, it will tend produce a better summary of the state of mankind’s knowledge.

    I have decided to judge these collaborations by this standard.

  12. mustacheion says:

    I can’t remember how I felt about the adversarial collaborations last year, but I personally was rather disappointed in most of the entries this year.

    Circumcision: 7/10.
    I voted for this one, because it felt like the only entry that was remotely useful for an individual, and because I thought it was truly an adversarial collaboration which a reader on either side of the debate could accept. I am a circumcised male and this entry somewhat reinforced my belief that circumcision has a small positive effect, with some severe but rare complications. But it also left me rather interested in the phenomenal experience of having an uncircumcised penis which is something I will never be able to experience and so never able to compare, but is still a relevant concern for this topic. I can empathize with the people in the comments who had a weak opinion on circumcision, no matter whether their opinion was weakly positive, or weakly negative. However, I found the opinions and experiences of commenters who were strongly against circumcision to be utterly baffling.

    Meat eating: 3/10.
    I can’t believe that the pro-meat eater actually held the position that farming animals is a net good because it maximizes utility for the animals, who would prefer mediocre existence to non existence. In my opinion this is making some very serious mistakes about philosophizing about non-existent minds, but I lack the will to try to further elucidate or defend this assertion. IMO eating meat is clearly a net harm, but not the largest net harm humanity is currently exposed to. Animals are tasty, food is an important part of my well-being, and I don’t have the slack in my life be able to make the choice not to eat animals without taking a serious hit to my well-being. But I am not happy about that, and would prefer it to be otherwise.

    Calorie restriction: 6/10.
    I liked the casual tone of this one, which made it entertaining and easy to read. Everybody seems to agree that the evidence is good in animals and poor in humans. It is very clear to me that for me personally, the gain to well-being from living a longer life would not be offset by the loss of well-being incurred by a calorie restricted diet. But perhaps others would make a different personal choice?

    Space colonization: 3/10.
    I read this one through, though I don’t remember it very well. I think I was annoyed by the pure lack of usable data. This question depends so greatly on the costs of actually accomplishing these tasks, and we know so little about those real costs, that I think it is silly to even pretend we can answer this question.

    Gene editing: 4/10.
    I think human genome editing is very dangerous, should be undertaken with great care, and is extremely important to our continued survival. But I understand the concerns of those who do not want this to happen. I think this entry did an ok job providing lots of information on how gene editing works and how it has been used so far, but did a poor job of framing the actual ethical questions. So, decent as a review article but poor as an adversarial collaboration.

    Abortion: 2/10.
    I think the philosophical framework used in this entry was simply seriously misguided on both sides. Contained in my opinion completely non-nonsensical conclusions. Personal priors on the question: abortion should be safe, legal, and rare. It is so much better to elide the messy question altogether by directing much greater resources toward preventing unwanted pregnancy in the first place.

    Robots taking our jobs: 3/10.
    I disagree too much with the fundamental beliefs of the economists whose work this was based on to have much regard for this entry. It didn’t feel very adversarial, and it didn’t feel like it had much of a conclusion either. I feel like this discussion wasn’t very useful; to actually answer this question, you have to discuss the nitty-gritty of the capabilities of real AI systems in much greater detail than this, and frankly at a level of detail that really nobody alive is qualified to speak at right now.

    Spirituality: TLDR/10.
    This was clearly the most professional entry. On the other hand, I grew out of thinking that writing like this actually contains information in college.

    My overall opinion is that the SSC community should not waste time or money on an Adversarial Collaboration again. I think this was a really cool idea when it was proposed, but I don’t think it is working very well in practice, and I don’t have a good suggestion for what to do to make it better.

    If I could have voted for the Grinch I would have.

    • JN340 says:

      “…frankly at a level of detail that really nobody alive is qualified to speak at right now.”

      I felt this way about the gene editing AC, having some expertise in the area. Too speculative. Light on actual science. No substantive debate, although that’s not the authors’ fault as such, in that the only people who could have a substantive debate about such cutting edge technology are people at the very forefront of it. I certainly couldn’t. Tough topic choice. Interesting essay if you don’t know much about CRISPR but I feel missed the point of an AC.

    • JN340 says:

      In fact, this seemed to be a theme this year. Space exploration to mitigate X risk, will robots take our jobs – who knows?? No one on earth is really qualified to answer these questions so any “argument” about them is going to be largely speculative. Ends up being more an essay than a debate.

      The ones I liked were calorie restriction, meat eating, and circumcision. Honourable mention to the spirituality one as well.

    • JN340 says:

      That said I want to thank all the collaborators for putting in an enormous amount of time and effort to educate us (and themselves) on various interesting topics. You’re the ones out there contributing to the community – I’m sure you’re not doing it for the money – it doesn’t pay very well! I learned a lot from all comers, even the ones that didn’t blow me away.

      • mustacheion says:

        Yeah, despite the negativity in my comment, I too appreciate the effort that went into these entries. I am critical only because I have high standards for this community and I think we can do better.

    • hnau says:

      I can’t believe that the pro-meat eater actually held the position that farming animals is a net good because it maximizes utility for the animals, who would prefer mediocre existence to non existence.

      I don’t agree with this position either, but it doesn’t appear to make much difference in their final utility analysis (and other approaches to constructing utility have other serious issues).

    • Shpoon says:

      I think I’d agree with most of these assessments – I voted for the circumcision piece since it moved my priors and presented what seemed to be all the relevant evidence. Spiritual Experience was my second choice in terms of how well put-together it was, but I also felt that it was kind of devoid of information that added value outside of the core question (I did find almost all the cases discussed interesting, and even familiar cases were discussed in ways I hadn’t heard before).

      I also have to echo the issues with the meat-eating piece – I’m sure the authors put forth a valiant effort, but the end product was very confusing and people on both sides of the issue who I sent it on to took issue with it for various reasons. As someone who thinks anti-natalist arguments often have logic on their side (even if they challenge many basic and powerful intuitions) I found the particular claim about the utility of simply existing kind of shocking. I usually presume that in order for a creature to prefer existing to not existing, it must have some faculties for abstraction that seem out of reach for cows, pigs and even dolphins (although suicide in the animal kingdom may suggest some creatures can actually express dispreference for life as well as preference).

      The research about pain and fish, as well as the neuroscience around cerebral cortexes etc, was fascinating though and I’m grateful to the authors for exposing me to it at the very least. My thanks to the other authors as well for their pieces, even though I don’t wish to comment on theirs at length.

    • Viliam says:

      I voted for this one, because it felt like the only entry that was remotely useful for an individual, and because I thought it was truly an adversarial collaboration which a reader on either side of the debate could accept.

      I assume you haven’t read the comment section — the only part of the article where the other side of the debate posted their actual arguments.

      Prediction: people supporting circumcision will claim that the debate was balanced, people opposing circumcision will claim that the debate was unbalanced.

      • NoRandomWalk says:

        Note: If you read the comments and still feel that way, apologies for talking to a strawman

        2nd’ing what @Viliam says, I would recommend reading the comments, especially the long one at the end by some medical professional with expertise.
        My takeaway is that the benefits of circumcision aren’t big when converted to life-years saved, it’s just a few very rare conditions (well, and HIV but if in developed world not an issue) whose risk is decreased further; and that in terms of sensitivity there could be a significant decrease in terms of pleasure from masturbation (maybe sex as well, but am less sure).

        Personally, engaging with the material (1st time I as a circumcised male ever thought about the issue) made me less against FGM on philosophical principle (I know exactly 0 about health effects, different types of it, etc) and probably more willing not to circumcise my sons if it became a lot less socially common. I.e. I’d want them to grow up as what is normal in their society for psychological reasons, as I think that is much more important than the physical impact.

        To be clear: I’m not saying it was unbalanced (having not read their writeup of the experience, don’t have enough information), just that it could have been even better with more context, and I am glad that the comments filled in some of that context.

  13. Rana Dexsin says:

    Voted! Thanks for hosting this.

    I too am curious what if any mechanism you have in place to mitigate ballot stuffing.

  14. Bellum Gallicum says:


    I felt the others lacked truly diverse viewpoints, felt more like two chefs deciding how to split the pig.

  15. Randy M says:

    I enjoyed the series and find it difficult to decide.

  16. blacktrance says:

    Ultimately voted for the meat-eating one, but I wanted to mention that the calorie-restriction one was also a strong contender, so I hope they get some of the prize money too.

    Also, how can you advocate for approval voting when you don’t even let us approval vote on these entries?

  17. meh says:

    Can I volunteer to do an ACC next year on the best way to vote for an ACC?

    • Robert Jones says:

      Probably better to do something more general, like “Should approval voting [or whatever] be used for most decisions, including ACCs?”

  18. Ninety-Three says:

    Scott is clearly putting his finger on the scale by keeping the Grinch off the ticket.

  19. caryatis says:

    In voting, consider: 1) is the entry high-quality? 2) is it truly adversarial, i.e. a person with a strongly held and well-thought-out opinion of X, along with a person with a similarly strongly held opinion of the opposite of X, and 3) is the question one capable of being answered, at least preliminarily, in this process?

    Your assessment may vary, but I saw several fairly high-quality contributions that were not truly adversarial. To be fair, it’s probably hard to find a collaborator for this process.

    • The Pachyderminator says:

      The final product isn’t supposed to look adversarial. It’s supposed to be a consensus report of a jointly conducted empirical investigation. If it looks like the record of a debate, which a couple of the entries did, then something is wrong.

      • JohnBuridan says:


        It seems to me that if a commenter’s personal position is not given enough space in the essay, then commenters assume that the adversarial element didn’t exist strongly enough. Scott explicitly asks for the essay to be a united front. We can only really know the level of adversarialness when authors write a retrospective on their process…

  20. Jeremiah says:

    For those who are still trying to get through them. They’re all available in audio format on the podcast feed. If that makes it easier.

  21. williamgr says:

    When is the (approximate) deadline for voting? (I still haven’t read all the entries, and don’t know how much time I have…)

    Also, (this is too late for this vote, as it has started, but not for future ones) why not use Approval Voting or (even better) Score Voting, instead of “First Past the Post”? Both of them can be easily implemented in Google Forms (using checkboxes for approval voting, or having a linear scale for each entry for Score Voting). Both of them are better than FPTP both in the case of “honest” and “dishonest” (tactical) voters.

    See also:

    To decrease the negativity in this comment: thanks for organising the ACC!

  22. Thanks for the contest! A lot of very interesting entries. 🙂

    Heads-up, though: There doesn’t seem to be anything in place to prevent me from voting more than once. Are you accounting for that?

    • NoRandomWalk says:

      As an anecdote, in an AP Psych class at a competitive high school I was in we all wrote 1 or 2 on a folded sheet and gave to teacher. If all picked 1, we get 1 point extra credit, if exactly one person wrote 2, only they got 2. I was the single person to write 2, not because I wanted the points but because it was the ‘game theoretic dominant strategy’ and ‘what a smart person would recognize as the thing to do, and since surely at least two would then the satisfaction of acting intelligent wouldn’t cost anyone anything.’

      I don’t think this is falsifiable, but I predict that everyone who votes will vote exactly once.

      • I do agree that’s what this community is most likely to do – although keep in mind it only takes one person to think otherwise to arbitrarily skew the results. Further, Slate Star Codex is not just its visible community; there are people who like to wreck Scott’s shit just because he’s Scott and has written things they disapprove of. I grant the worst that can happen in this particular case is that an “unpopular” adversarial collaboration gets the money, but I am still automatically concerned when I notice what I did. 🙂 (I am reasonably sure entries I didn’t vote for will get first and second place, regardless of whether trolling happens or not, for the record.)

        It should be mostly fine as long as he can look at timestamps (IIRC Google Forms automatically retains timestamps), which at least rules out the most lazy trolling, should it happen – which is why I ask “are you accounting for that?” rather than “please change that!”.

      • Ralf says:

        How did your class mates react?

        • NoRandomWalk says:

          It was right at the end of class and responses were anonymous; I don’t remember (was some years ago). I think one of three classes had everyone ‘cooperate’. I don’t know if the other class with a defector had exactly one or more.

      • Radu Floricica says:

        But… it’s not the thing to do 🙂 I once went down the rabbit’s hole with a writeup by EY on Timeless Decision Theory (Newcomb two box problem etc). Long story short, you have limited control over a collective because you are a member of that collective – there is a decent chance that whatever thought process you use to reach your decision is also used by others, including this particular strategy. So you can chose to cooperate – and expect a very weird (a)causal effect because others will also chose to cooperate using the same logic.

        • Milo Minderbinder says:

          I think about EY’s Newcomb writeup all the time. Thinking that my decisions generalize to individuals sufficiently like myself has made me more generous/nicer, since I’d like people who are like me to be like that.

      • Dacyn says:

        ‘game theoretic dominant strategy’ and ‘what a smart person would recognize as the thing to do […]’

        Yeah, an unfortunate common effect of teaching people game theory seems to be to make them conflate these two things…

      • fortybot says:

        The framing of the question matters a lot as well. In a class I was recently in, quizzes were curved so that the top score on a quiz was 100%. Now, the collaborative strategy here is for everyone to mark down “C” on every question so that everyone gets 100%. However, no one ever tried to collaborate on this, and, according the the instructor, no one ever has.

        • Robert Jones says:

          Surely marking down “C” isn’t a Schelling point: why not “A” or “B”? So you would need to communicate to agree this strategy which you can’t do, since you would be exposed as cheaters. Also, it’s not analogous, because the person who is able to get the top score gets 100% by defecting, regardless of how many other people defect.

          • The Pachyderminator says:

            I remember hearing that C is the most common correct answer on four-option multiple choice tests (when the test is written without the aid of a randomizer, which probably isn’t the case for savvy teachers). In much the same way, asking people to pick a number from 1 to 10 usually yields 7.

          • fortybot says:

            The specific answer doesn’t matter. And I don’t think collaborating would be cheating; if I tell everyone to mark down “C” for every answer before even seeing the test, there’s no way I could be divulging answers.

            > Also, it’s not analogous, because the person who is able to get the top score gets 100% by defecting, regardless of how many other people defect.

            Well, they may get less than 100% (if they defect poorly). Of course, if two people defect, then you have the same situation as before the test, which may not be favorable. In this class, for example, the average score was probably around 50-75% pre-curve. But yes, this is not an analogous situation.

      • eric23 says:

        Writing 2 is the best strategy if you care about your own grade and are completely indifferent to everyone else’s grades. If you care somewhat about other people’s (for example: your friends’) grades, writing 1 is the best strategy.