I see no reason to expect that popular voting will lead to the best argument winning out for issues where a lot of evidence has to be understood and set in relation.
It seems to me a pretty strange decision to want the barrier of entry as low as possible by allowing IP editors.
I see no reason to expect that popular voting will lead to the best argument winning out for issues where a lot of evidence has to be understood and set in relation.
Does this mean you think the idea, at root, is not worth it, or that you think it will help with some issues and not with others?
It seems to me a pretty strange decision to want the barrier of entry as low as possible by allowing IP editors.
I expect that a high(er) barrier to entry will produce a self-selected subpopulation that will sometimes miss out on important ideas or points that people outside that subpopulation would have thought of. I'm willing to put up with a great deal of dross in order to make sure that all the good stuff is caught.
Epistemic status: I'm not the first person to think of this
I think one reasonable way forward would be to compile a list of as many previous attempts along the same lines as you can find, and then analyze what about them worked and what failed.
It's my impression that there have been a lot of attempts.
Right off the top of my head, every debate website I've come across so far puts topics into simplified yes/no questions instead of considering multiple possible alternatives next to each other. That's true of Kialo, DebateIsland, Debate.org, ProCon.org, CreateDebate, iDebate.org, and more.
That's a really cool site. I think it only can cover truth-claims from the past (not proposed actions or if-then truth-claims about the future), but it will really excel at those. I'll keep it bookmarked.
they said somewhere (don't remember where) that they aim to bring it to that level, but due to the in-activeness i suspect it won't ever get there... which is unfortunate...
I like the broad vision imagining as many useful tools as possible to throw at the problem, but this has too many parts competing for attention, at least for a product without a mature community. What do you think is the minimum subset to build and sustain a userbase?
A few other thoughts:
Most issues don't actually have just 2 sides and grouping subtly different points of view into one side leads to apparently inconsistent positions which are frustrating to argue against.
I don't think reaching consensus is generally possible for the kind of arguments you're interested in, even with the aid of new technology. We know technology can help speed up processes we already engage in, but you're hoping for a different outcome to debates that have been going on for years or centuries. How do you make users feel that participating is worthwhile when actually convincing the world is unlikely?
I like the Open Forum. I was considering a similar idea (to be named something like "Manifesto") years ago that I think of as Pinterest for arguments. The goal wasn't to achieve consensus or "solve" the issue but for each participant to have a public and well organized position on any debate or sub-debate that interested them. You organize the arguments hierarchically and link-to or crib-from-and-edit other people's points or expound a new one. The key is everyone has their own argument tree, so an expectation of consistency is reasonable, but all the effort is in improving or adding new thoughts. The value to a user then is first, having a record of their own beliefs and development, and second, the ability to diff another tree and quickly find the heart of a disagreement or new insights to incorporate. And unlike a dialogue, if someone does convince you, you can quietly acknowledge your changed mind without eating crow.
What do you think is the minimum subset to build and sustain a userbase?
Really large, which is a major fail point.
I don't think reaching consensus is generally possible for the kind of arguments you're interested in
I think consensus is not possible for some of them; we're not going to "solve" abortion or God. On issues like that, the best that could be accomplished is helping people understand where the other is coming from and reducing animosity a little. (Which I think would be very worthwhile, if that could be accomplished, but even that might not happen.) Some compromise actions might possibly become popular, like privately funding programs for low-income women who might otherwise have abortions.
On other topics, I think we might be able to come much closer to a consensus than we are. Maybe not 100%, but a well-laid-out argument for adopting a different voting system, enacting a particular set of campaign finance reforms, a phase-out plan for eliminating or changing government farm subsidies, or a suggestion for how amazon can increase employee satisfaction without losing profits...those might make it pretty close to consensus.
Pinterest for arguments
That's...a really interesting idea. That might satisfy my desire to quickly find all the important aspects on an issue in one place. You'd have to mentally build the organization between ideas and options yourself, instead of having them visually laid out, but you'd spare yourself the trouble of forcing people to build or agree on that organization. Are you going to build it? Do you think a lot of people would use it if you did?
Yes. It's better than the alternatives I've seen, but it still feels seriously insufficient to me. Some of that is just because Kialo itself isn't large/popular enough to have comprehensive points made on it yet. But my bigger objection is that it feels simplistic. Example. Kialo presents: "Should Governments Ever Limit Free Speech?" with a series of mostly one-sentence points on either side. It doesn't examine different possible ways that governments have or could limit free speech, and the possible or real-life past consequences of each. The arguments on both sides are almost exclusively value-statements with almost no reference to possible facts. (Values should, generally, dominate pro/con arguments about actions, but facts should be appealed to in order to support the claim that an action does/does-not support a value, and more so when claiming expected consequences.)
Kialo doesn't offer a way to break down the question into more specific options (maybe real free speech shouldn't be limited, but political contributions should stop being counted as speech; maybe Germany should limit holocaust-denial speech but the U.S. shouldn't). Kialo's format currently encourages people to post opinions on topics more than it encourages people to think more deeply about topics. It's possible that could be changed with greater participation or with Kialo somehow working to create a culture of deeper thinking. But I don't see that there now.
I also think that logging in to post is a barrier to it growing as a crowd-sourced site. And I dislike showing the Like votes instead of only using them to sort, because I think that encourages people to pay attention to how popular an opinion is, and it makes those with minority opinions painfully aware that they are in the minority.
(Is minimum wage a good thing? Should I adopt a paleo or keto or vegan or Shangri-la diet? What do we really know and not know about the historical Jesus?)
I would point out that the three examples you've listed are of three different categories. The first, "Is minimum wage a good thing?" has a significant value component. Do you value whether people have money? How much inefficiency are you willing to trade off in the economy in order to ensure that people have a certain amount of minimum spending power from work? Without knowing your specific values, I cannot answer whether a minimum wage is or is not a good thing.
Your second question, "What kind of diet should I adopt?" has significant dependencies on your physiology. Are you gluten-allergic? Do you have allergies to nuts? Do you have diabetes? Kidney issues? All of these things impact which of the listed diets (if any) is going to be best for you. And this is just from a strictly physiological perspective -- it leaves aside issues of preferences (i.e. maybe veganism isn't really right for you if you really like bacon).
The third question, "What do we really know and not know about the historical Jesus?" is answered, to a first approximation, by Wikipedia.
I think you're really asking for three sites. For the first question, there should be a site where people can debate moral values. Ideally, this site would taboo "good" and "bad" altogether, and force people to frame value judgments in the context of the value systems that result in those judgments, allowing others to see the criteria by which those judgments are made.
For the second question, a site that provides guidelines rather than recommendations would be helpful. Ideally this site would present a way for you to submit details about what your medical situation is and what your dietary preferences are and then it would output a decision tree that you could use to arrive at a diet that would work best for you.
Finally, for the third site, it'd be something like Wikipedia (only perhaps with better filtering tools to help weed out out unsourced data).
I'm not sure that it's possible to put together one site to rule them all because the the they're doing such different things. We're going from "there might not even be a 'right' answer" to "there is a right answer but it might be different for every person" to "there is a single, externally verifiable objective truth". How do you handle that range of epistemologies with a single site?
Minimum wage is actually somewhat like diet, since it could be that some places and not others would be better off adopting it, depending on their varied conditions. While values dominate discussions of actions, I think the epistemic questions of what the consequences of those actions are are very important. And "if X, then Y" is a claim of truth.
In the end, I think that both actions and truth-claims rely heavily on both objective truth and on values. Valuing Breitbart or Slate as a reliable source can determine what facts you believe, and it isn't possible to completely divorce questions of truth from values about sources, about what level of evidence is needed before accepting a claim, and such. I would like to make those more explicit.
I do think the diet thing would be the kind of question that would be hardest to succeed at. I think the Site could handle some degree of "different solutions for different situations", but the level of variability in medical questions might be beyond it.
Someone claimed that https://www.metaculus.com actually has a pretty good prediction track record
Tim Berners Lee was, some time ago, working on the idea of web 2 which would be in accord with your idea, since every web page metadata would include the possibility of being coordinated across websites, categorised, authored, timed. As I understood it there would be cross referencing and precision to searches and, unless I am very mistaken, would be in agreement with you that such things would be possible given time and money and knowledge. Tim berners lee could not write his ideas in one simple reply to a thread so all I can do is tell you to visit his website and read on what he is working on, as he is very interested in ideas such as yours. I am absolutely certain that he, or his team, would be very open to your ideas and may even suggest possibilities you had not thought of
Hi Anna, I thoroughly enjoyed your article. My team and I have been working on a project that attempts to solve some of the problems you've outlined. I'd love to get your thoughts if you've got time to check it out:
https://letter.wiki/conversations
Best,
Clyde
Is the Letters site itself, the project you mention, or was one particular conversation on that page discussing the idea (if so, which one)?
Epistemic status: worried about effort/time lost.
I am by no means experienced with any of this, and seriously considered not writing anything at all. But it only takes me a bit of time (an hour max) to write why I feel why the odds are very strongly against you, and if you are serious about pursuing this idea then even if at low probability that my comment is helpful to you it's worth writing it on average. So here we go.
During my read of the post, top-to-bottom, at the part
On matters of truth, it needs to support epistemic arguments for why we should believe or not believe particular claims. On matters of action, it needs to provide important pro/cons of taking that action. Site must have a method of allowing the best arguments to rise to the top.
my internal monologue went "The first bit is difficult but perhaps possible. The second is a mess. Oh dear, the third is basically impossible!". The sentence immediately after, explaining this functionality would be the bare basics, shocked me quite a lot. I think aiming for the quoted section is nigh-impossible, and then we haven't started on the possible additional features you mention. Your post strongly reminds me of Benjamin Hoffman's piece on Anglerfish (in my opinion worth reading in full), and also a bit of a segment (near the start) in one of Eliezer's posts on security mindset - where the character Amber makes the mistake of thinking that the critical part of her startup is the technology, where really it is the security. I think in a similar manner your Site would, besides depending on the UI, the back-end, the marketing etc. also depend critically on its ability to continue growing during certain critical phases, and the lack of discussion on this as a plausible failure mode is making me rather pessimistic.
In my mind, conditional on Site eventually operating as intended, it should grow through several phases. First you have a low number of users (~100 regular users? Sorry, I don't have experience with this) who basically filtered in from your social circles, and are able to aggregate their opinions/thoughts as intended. Then in the next phase Site grows more popular as people notice this is a valuable source of truth/plans/speculation, and they provide new questions and answers covering broader topics. After that there should be some third phase where Site is diverse and big enough that all those extra features you mentioned might become plausible to implement (I'll come back to this later).
My problem lies with the second phase. Benjamin's piece suggests that as soon as Site is big enough to have any real value, this immediately creates incentives for outsiders to try to abuse/free-ride on the project (for example through manipulating the questions or voting). This would be worse on discussions on *actions*, which is why at the start I mentioned that that is more difficult than discussing *truth*. Your wish to keep Site crowd-sourced makes it more difficult to guard against this phenomenon, and to me Eliezer's writing on security mindset suggests that if you don't treat this problem as central the odds are strongly against you. It is unclear to me what motivates people to keep coming back to Site in this second phase if they disagree with a large part of the demographic/consensus, or in general why echo-chamber effects would not apply. In fact, it is unclear to me why people would spend time participating in discussions outside their immediate interests at all (see also for example evaporative cooling).
Lastly I think a large part of Site would only function after you have some critical mass of users to have sufficient discussion on a lot of different topics. This is troubling as it means those parts existing at all is conditional on Site being a success. In the spirit of "If you're not growing you're shrinking" I think a lot more time and effort should be focused on figuring out how to obtain and keep a userbase, and introducing fancy features is downstream from this.
Sorry for being so critical and nonconstructive. I don't know how to solve any of these problems, but like I said at the start it felt like a wrong strategy to just stay quiet. I hope I'm wrong about most/all of this, and let me as a closure mention again that I don't have experience with this at all.
Those are all concerns I share. I don't have solutions either. I feel like my choice is to either build the website despite the lack of solutions and the high risk—or settle for not having anything that does what I want.
If I tried to do research on how to make websites grow, I would expect to encounter a lot of advice that's based on survivorship bias, and therefore unreliable. (I mostly expect that luck is a/the dominant factor.) Do you think research on that would produce helpful results?
Moderation, on the other hand, is probably something that I could start with some research on, to see what might or might not be possible/helpful.
Well, I still don't have any experience with this. But maybe possible avenues include:
The survivorship bias is a very valid point, but [not doing research on how to make websites grow] is also a poor strategy. Personally I'd still look into the advice, but I'm afraid what you're trying to do is simply very difficult.
Hey uh, I've been thinking all of those thoughts too. We should probably nucleate up a community (a slack channel or something, somewhere to hang out and share our findings and make plans) because I'm pretty sure there are at least 10 people knocking around just here who have their heads as far into this as we do. Heard Eliezer was absolutely overflowing with discursive technologies when Arbital was being planned, his concepts were fractaline. I've been that way. I guess I pulled back a bit when I started to understand that having infinite visions of sophisticated collective intelligence augmentation systems isn't really the hard part, the hard part is building any of it, funding it and holding users.
I do see some ways to do those parts.
I'm just gonna start talking excitedly about the most recent piece of the puzzle I turned up because until this moment I have not had many people to talk to about this (lots of friends who're interested but not many who'd ever take what I was saying and do anything with it)
Yesterday I flipped out a little when I remembered that article Scott Aaronson did about eigenmorality (eigentrust) and I realised it this is exactly the algorithm that I've been looking for months, for doing a basic implementation of the thing you're calling "Contrast Voting"... (I'm going to keep calling it order-voting and graph rank recovery if you don't mind? Idk I think there are more standard terms than that) I haven't tried it yet (I just found it yesterday. Also I want to port it to Rust) but I'm pretty sure it'll do it. Basically what we need to implement order voting is, we need a way of taking a whole lot of order pair judgements/partial rankings from different users and combining them together into a single global ranking. With eigentrust (similar to all the other stuff I've been trying), basically what we'll do is we'll build a network graph where each edge represents the sum of user judgements over the two candidates, then we run eigentrust on the graph (it's a similar technology to pagerank, if you've ever had that explained to you. Score flows along directed links until an equilibrium is reached), and then we have an overall ranking of the candidates. We'll need to do some special extra stuff to allow the system to notice when two clusters of candidates aren't really comparable with the data we have, and it'll probably need to try to recognise voting cabals because there's a dark horse problem where-...
I should really write this out properly somewhere.
The reason I haven't done that already is that I'm not sure how many of our concepts should be exposed publicly.
these technologies are actually powerful. Even just order voting alone would speed up content sorting by like 20x, imgur could use it for recommending fucking cat pictures and they would become even more compulsive than they already are. (They might already be using it in a hidden way, I think netflix is.) Power isn't good or bad on its own, but some powers are more likely to be put to good uses than bad. Collective intelligence platforms are more likely to be put towards good uses than AGI is, they're inherently made of humans, they're more likely to reflect roughly human values even when they go wrong, but in their worst incarnations they can still just end up becoming completely insane demonic egregores like... dare I even speak their names, no, no I daren't, because I don't want to draw their millions of eyes towards me. Let's just say that some of the social media platforms I frequent most often are basically incapable of forming sound epistemic structures, and I'm afraid of most of their segments there, and I really hope that those words they're saying never become much more than words.
The ideas I have for technologies that'd gather and harmonise users quickly and efficiently are also some of the ones that scare me the most. I know how to summon an egregore, but Making the egregore come out of the portal sane takes a special extra step. It's absolutely doable, but I wouldn't trust anyone who's not at least weirdsun adjacent to understand the problems well enough and to stop and think about what they're doing and put in the work to make it all turn out human, and maybe not release it onto the internet before it's sound.
I think the first step is to make something that gathers information that people want. A place where people will feel comfortable forming communities and spending time. A humane place, something that respects peoples' attention, rewards it.
The world needs platforms where good mass discourses can exist, currently we have, actually none.
I actually think this should be an EA cause. At some point, if we can gather a decent team, we should start asking for funding. Maybe move to the EA hotel in blackpool and grind on it for a bit once we have a 1.0 vision.
I...only followed some of what you said here. *Googles slack channel* ... Sure, if you know other people who are interested in a similar concept, that might be worthwhile. How do we go about it?
What is weirdsun?
I'm very excited about what might happen if we got ten people like us in a channel, I think that's a community/project I'd give a lot of energy to, but that didn't occur to me until just partway through reading your post, so I have not been collecting any names until this point, sorry. Maybe we should wait til we have a few more than two, before I start sending out invites (by the time we do, there might be something nicer for async group chats than slack).
(weirdsuns are... analytic surrealists. I don't know if I'd say they're influential, but as a name for a certain kind of thinker, those unmoored by their artificial logics from the complacency of common sense, they're a good anchor on which to ground a label.)
Epistemic status: I'm not the first person to think of this, and I'm not sure if this is possible. But I want to flesh out some options for consideration and get feedback. Sorry if it's long.
WANTED:
A crowd-sourced site that reliably presents important pros and cons on any/all topics and hopefully leads us collectively closer to consensus.
This has been on my mind for a few years now. Figuring out truth in the digital age is often tricky and sometimes downright impossible. (Is minimum wage a good thing? Should I adopt a paleo or keto or vegan or Shangri-la diet? What do we really know and not know about the historical Jesus?) Especially when it comes to political discussion, answers often depend on more details than an average person has time or energy to research—and even if they did, they may still leave out considerations or solutions that someone else may think of.
Some links where other rationalists think semi-related things: Robin Hanson wants AI to help us towards consensus because Wikipedia isn't up for the job. Scott Alexander promoted adversarial collaborations to see if they could reach consensus. (I swear, I saw more examples of people wishing for something like a debate website, but I wasn't writing them down and searching for them has failed.)
THE WISHFUL THINKING PART
Now let me be more specific about what my imagined website (hereafter, Site) needs to do.
On matters of truth, it needs to support epistemic arguments for why we should believe or not believe particular claims. On matters of action, it needs to provide important pro/cons of taking that action. Site must have a method of allowing the best arguments to rise to the top. That's the most basic functionality.
Additional Feature A: Because the Site is crowd-sourced, anyone can propose possible actions and possible modifications to actions. In theory, this could mean that Site could be a platform where a crowd-written law could have its details hammered out by many affected parties. Creative solutions could have a chance to be reviewed and, if well-thought-out, could rise higher in the public consciousness.
Optional Feature B: I really, really would like it if, within the structure of arguments (for either epistemic-truth topics or proposed-actions), Site distinguished between truth-claims and value-claims. This might be especially unrealistic, since no one naturally lays out arguments that way. But gosh dang it, I really want it to.
Optional Feature C: A number of rationalist posts have discussed how useful it would be for society to have a way to coordinate on certain problems: specifically, having a conditional "I will adopt X if Y% of other relevant people also do". It might be possible to set up Site so that proposed actions that need a minimum number of people to adopt it in order for it to be worthwhile offer the option of conditionally committing to adopt it and sending out notifications when thresholds are reached.
THE NITTY-GRITTY
Obviously, this is a LOT to ask from a website. Before I go into everything that could go wrong and all the reasons it might not even be possible, let me lay out a couple of operational options.
Contrast Voting
Reddit and other up/down-vote systems suffer from certain effects, an important one being that late-to-the-game comments don't stand a chance against early popular comments, even if the later ones are higher quality. To combat this, Site should not allow up/down votes on individual items/comments/whatever. Instead, Site should present two items and ask for a vote on which one is better. In this way, Site can give better rankings to items. A new argument, perhaps one that presents a new, more recent study finding, can quickly outrank the previous top argument if, when those two arguments are presented, most people vote that the new one is better.
Contrast Topics
The Contrast Topics option would not have a page titled "Is God real?" nor a page for "We should outlaw chicken cage farming." Instead, topics would be presented in terms of possible alternatives. "The Abrahamic God is real" could be separately contrasted with "There is no God" and, on another page, contrasted with "There is a divine energy". "We should outlaw chicken cage farming" could be contrasted with "We should eat less chicken and let the market adjust chicken cage farming downward" or "We should slowly phase out chicken cage farming according to this detailed plan I've written out" or "We should wait until chicken-like lab meat is available, and then outlaw chicken cage farming".
Visually, this might be tricky, but a "Chicken cage farming" page might list all of the proposed actions related to it, and a user could select two actions and click "next" to see the pros and cons of those two actions relative to each other. (These would be more like pros of A vs pros of B, not pro/cons of each; a con of B would be listed as a pro of A and vice versa.) These headings might be treated as tags/labels rather than a strict hierarchy of categories, so that a possible action or topic might appear under multiple headings if it's relevant to all of them. Headings might, in turn, have both epistemic truth-claims and potential actions listed under them.
Open Forum
In this option, individual users compose complete arguments, either pro or con. They are allowed and even encouraged to steal what others have written, add/subtract/modify it as they like, and submit it as a new argument. Contrast Voting is used within "pro" and within "con" to rank these arguments and present the highest ranked at the top. Probably only the top one or two submissions will show by default (the rest reached by an expandable + or "show more" or such), since submissions are expected to (eventually) aggregate all the best arguments.
Wikibate
This is an alternative to Open Forum. It works more like Wikipedia; everyone is allowed to directly edit one single page. A pro and con side are provided, and we see if a version comes out that makes enough people happy enough to be stable. A behind-the-scenes contrast voting option might be used to decide which order the possible arguments appear in, but the phrasing of the arguments would have to satisfy people or be edited by someone.
Personal Expertise Stories
In normal life, people rely heavily on personal experiences when forming their beliefs. If Site doesn't sometimes take this into account, it will feel shallow to many people. Argue the facts of abortion all you want, but if you can't include someone's vivid description of having a loved one die from a back-alley abortion or another person's horror at seeing what they were told is a "lump of tissue" that turns out to look exactly like a baby, you're missing important context.
It is also the case that sometimes hearing expert testimony is helpful. Lawmakers often rely on it. Experts don't always agree, but a claim from an expert can and should carry a different weight than the same claim from a non-expert.
On a crowd-sourced site, anyone can claim they're an expert or make up b.s. stories about what happened to them. A possible solution is to allow people to submit a "personal expertise/story". If submitting one, you have to include personal contact info like a phone number. Some number, say 10, people can apply to verify a story. Only approved verifiers are given access to the story submitter's contact info. The verifiers promise to come back and report on the status of their verification and not to abuse the contact info. (A complaint from a story submitter results in that verifier being unable to do any further verifications, and possibly banned from the Site.) The verifier calls the person and talk to them, get details if possible, like other people or companies they can talk to, to verify parts of the story or claim. In the end they report back, maybe on a 5-point scale, whether they believe the person's claim, and the net verification result from all the verifiers is displayed with the story. (E.g. a green bar showing 3.5/5, with a (2) next to it indicating only 2 verifiers have submitted responses.)
Consequences Before Arguments
For proposed actions, before the actual pro/con arguments appear, I would like to see a list of possible consequences that might result from taking that action. While I don't want Site to adopt consequentialism as an official philosophy, I remember (Google, you fail me) an article that claimed that people demonstrated less partisanship when they were asked to think about the consequences of possible laws. A potential consequence is itself a truth-claim ("If we do X, then Y will happen"), so that should link to its own page.
Login Voting
Generally, I want people not to have to create accounts in order to participate on the website. IP addresses can be recorded, like Wikipedia does, for submitting/modifying arguments, proposing actions, and creating topics. That should minimize the barrier to entry and encourage participation. However, all voting should require logging in, to minimize one person voting multiple times, and to accurately track if someone changes their relative rankings of two arguments.
Parallel Axes Voting
There is no one standard for what makes an argument "better" than another. Relevant measurements include: accuracy, importance/applicability, thoroughness, kindness, formatting/grammar/spelling. One option might be to allow users to Contrast Vote two arguments in multiple categories separately: Argument A has more important points than B, but B has better grammar than A, and they are the same on kindness. Parallel Axes Voting is more relevant if Site uses the Open Forum style, and not so much if it uses Wikibate.
THE PROBLEMS, OH SO MANY PROBLEMS
In the end, despite the problems, I think Site would be more worth having than not.
So what do you think? Is Site even possible? Is there a better setup than Open Forum or Wikibate? If not, which of those two is better? Should the Personal Expertise Stories option be included? What other details could Site implement to be successful? Are any of you willing to work on it? Is it worth my time to work on it?