This is another experiment in self-resolving markets. Some previous attempts:
This type has this issue that a single large trader can force the market to resolve incorretly and turn a profit in doing so.
This type has the issue that it may not resolve for a long time past the close date
This type has the issue that it's trying to exploit a cognitive bias among traders who would otherwise exploit the market, and isn't otherwise any more robust to manipulation than the first type.
And all three of those have the issue that they'll tend to stabalize a little below 100% or above 0%, since it's so cheap for a single trader to keep the probability from going all the way to one extreme.
This market is a new attempt with I hope will avoid the problems of the previous 3 types. I expect it to have its own new and exciting problems instead. :)
Upon market close, I will post a YES/NO poll in which each Manifold account can only vote once. The poll will not be anonymous; everyone will be able to see who voted for what. This market resolves to the majority result of the poll. (For example, if 49% of voters vote YES and 51% vote NO, this market resolves to NO.)
This works by:
Preventing someone with more mana from forcing the market to go where they want it to, by ensuring that every user gets an equal vote.
Using the threat of social dissaproval to keep people from voting for a result that's clearly "wrong".
But each of those has their own downsides that increase the chance of an incorrect resolution:
More accurate predictors don't have any more say over how this market resolves than less accurate predictors.
Social pressure isn't always in favor of the right answer. >50% of traders might believe that YES is the correct answer to a certain question, but without common knowledge of such, they may risk social penalties for voting YES and all choose to vote NO instead. Having the poll be anonymous would solve this problem, but increases the chance that people will coordinate to vote for an answer they know to be wrong.
1,000
3.00One thing to note: This polling system is subject to bias from the market creator, as I could have continued checking back and advertising the poll in new places until the votes lined up with with my desired resolution.
Future markets using this system should require the creator to precommit to a time when they'll check the results.
@IsaacKing I was considering using the like-a-comment poll (as opposed to e.g. a poll on an external site), but one difficulty is that votes could theoretically be changed after the time when the vote was supposed to close. As an author you either have to precommit to checking the results at the exact time, or risk votes changing later without you catching it. And theoretically if there's a burst of last-minute voting it could be hard to ensure the correct vote count.
Still, I think it's good enough for most purposes to specify a deadline and then make a best effort to check shortly thereafter.
Ok, here's the poll. Please vote by liking one of my two replies to this comment.
If you believe that Joe Biden was the president of the United States at the beginning of 2023, like the comment that says "YES". If you do not believe that to be true, like the comment that says "NO".
I will check the likes at an arbitrary point in the next few days and resolve the market based on the result. If the counts are tied at the time, I'll just check back later until it's not tied anymore.
@IsaacKing this is a very convoluted way to avoid triggering the controversial markets before end of Feb market.
@CromlynGames I'm confused, how so? This market isn't controversial and was also created before that market so it wouldn't count even if it was.
@jack I'm not arguing this OP market is potentially controversial. I'm talking about Isaac's poll comment, and getting people to Like one of two following comments to vote is basically a binary question about a controversial topic. I'm arguing that if he wasn't trying to avoid triggering his controversial-markets-market, he'd have done the poll as a separate market (as is common for a lot of subquestions). I believe the topic of who won the 2020 presidential election is still controversial for a good chunk of Americans? (most of the courses are from 2020 though) https://news.northeastern.edu/2020/12/11/who-won-the-2020-presidential-election-joe-biden-or-donald-trump-depends-whom-you-ask/
@CromlynGames It's a somewhat common way to resolve some types of questions - see e.g. https://manifold.markets/jack/poll-should-it-be-possible-to-dispu or https://manifold.markets/Yev/will-it-be-unclear-whether-destiny#8D05k4lqXqQDuwLcGuLm or https://manifold.markets/Sinclair/which-hogwarts-house-is-the-best-po
And it's literally what he said he would do in the description
Upon market close, I will post a YES/NO poll in which each Manifold account can only vote once. The poll will not be anonymous; everyone will be able to see who voted for what. This market resolves to the majority result of the poll. (For example, if 49% of voters vote YES and 51% vote NO, this market resolves to NO.)
@XComhghall I'm referring to two different types of social pressure, to be clear.
One is from Manifold users in favor of not resolving markets improperly. That's the type that will (hopefully) make this market resolve correctly.
The other is from general, non-truthseeking society, that pushes people into agreeing with their friends and coworkers. That's the type that I fear would make a market on a more controversial topic resolve incorrectly.
@XComhghall And you point out two important things:
With regards to the first type of social pressure, it doesn't work if manifold users don't see a NO vote in this poll as a reason to not trust you on this platform and not bet in your markets. I'll be interested to see whether you face any repercussions for your incorrect vote.
With regards to the second type of social pressure, pseudonymous accounts do help prevent that. But the more popular a certain account gets, the harder it becomes to keep it pseudonymous. Additionally, as soon as that persona starts doing other things, the pressure can apply to them even anonymously as well. For example, I could easily see an anonymous Twitter account losing a bunch of followers if they said something those followers found transgressive.
I was also thinking about point 1 - if when running the poll you explicitly state that voting for a resolution should be considered just like choosing a resolution for your own markets, then I'd expect the norms on improper resolution to work reasonably well. But I think if that isn't very clearly stated by either the author or the community, I wouldn't be surprised if people see it as just a poll that isn't tied to their author reputations then it doesn't work nearly as well.
And point 2 - if you restrict the poll to people who have established author reputations, then you avoid a lot of problems.
Point 1 - I don't think whether the creator of a market clearly states that is important. Voting in a poll as a trader is fundamentally different from resolving a market as the creator. As Jack (I think) previously said, the actual questions of resolving-to-market markets are known to have no implications, and may simply be ignored, and the results do not need to reflect the true opinions of traders, or their consensus, much less the truth. This is fundamentally different from markets resolved by the creator, where the creator is expected to resolve the market based on reliable, commonly accepted information, and be truthful.
@XComhghall I disagree. This is very much not a resolve-to-market question. In those questions, the author clearly states how they are going to resolve, and carries that out as specified - nobody is doing anything dishonest in terms of market resolution, and that is why there's no issue. On the other hand, if I ask someone else to choose a resolution for me, then there is an expectation that they resolve correctly.
There are cases where people delegate the choice of resolution to one or multiple other persons. If those people resolve dishonestly, that reflects badly on them in my eyes and also in the eyes of most of the community. A poll is just an extreme version of this where the resolution is delegated to the entire community. I do agree that in general there may not be a strong consensus on how much an incorrect vote in the poll weights on your reputation, but I think the author doing the delegation has substantial power to influence this - they can declare "this vote is just for fun, do whatever you feel like" or they can declare "vote as if you were resolving your own market".
@jack Hmmm . . . That is true. If the creator says, 'Vote as if you are resolving your own market', that does seem more serious.
Otherwise, I do think in general that delegating the resolution to anyone who would like to participate in a market or a poll on a website like Manifold has no implications on the traders or voters' true opinions or consensus.
It depends on both the creator's and the traders' expectations. Even if the creator says, 'Vote as if you are resolving your own market', are we expecting participants in the poll to vote truthfully, knowing that the author is resolving it through a poll open to anyone who would like to participate?
I do not have such an expectation, and would not blame anyone who treats the poll as just for fun.
For this market, I'm not going to provide voters with any instructions, since I didn't say up front that I would, and I'm curious to see how many choose to vote incorrectly. I will not consider an incorrect vote to be dishonorable in the same way as an incorrect market resolution.
I have created a duplicate of this market where I will consider incorrect poll votes to be dishonorable, to see if there's any notable difference in how people vote.
Also, one frequently asked question about markets-on-polls like this is, why not just run a poll? One big reason is that you can predict the result of a future poll (like in this market or https://manifold.markets/SneakySly/at-the-end-of-2023-will-manifold-us). Another reason is that you can use randomization to predict the result of a poll that you only run say 10% of the time.
Overall, I think this is a very useful mechanism. And it's fundamentally different from self-resolving markets because the resolution references something external to the market itself - namely, the poll.
The biggest concern I can see is how to prevent someone from manipulating the result by generating a million fake accounts (https://en.wikipedia.org/wiki/Sybil_attack). Manifold currently addresses this by requiring a google account. You could also imagine addressing it by requiring putting in some amount of mana to vote.
There's also a concern that polls aren't necessarily the best way to aggregate information, because everyone gets an equal vote regardless of what information they might have. That's basically the motivation for using prediction markets rather than just averaging people's predictions. But here we still get the benefit of the market for the prediction part, and we just use the poll for resolution.