This is based on /Tetraspace/will-the-responder-onebox-in-this-t but with modified amounts and with a market predicting the specific responder's decision.
Here I will run an instance of Newcomb's problem, a decision theory thought experiment. The setup is as follows:
There are two boxes designated A and B. The player is given a choice between taking only box B or taking both boxes A and B. There will be a Manifold prediction market which predicts whether the player takes just one box or both boxes.
Box A always contains M$1.
Box B contains M$1000 times the probability that Manifold predicts of the player taking only one box. (E.g. if the market predicts 30% chance of taking only one box, Box 2 contains Ṁ300)
Both boxes are transparent, i.e. the player knows exactly how much mana is contained inside them.
I will select a random participant in this market at close to be the player in the Newcomb's problem. After selecting them, I will create another market where Manifold predicts whether they will pick only one box. (The player is welcome to talk about their thinking, or to say nothing. They are not allowed to trade in the market.)
At the conclusion of that market, the amount in box B will be set, and the player will decide between taking only box B or both boxes. The player will receive a manalink for the amount of mana contained in the box or boxes they take.
This market resolves YES if they take only box B, and NO if they take both boxes.
Fine print:
If the player doesn't respond within 2 days of being selected to confirm that they are playing, or doesn't make their decision within 2 days, I will randomly choose another participant to replace them
The player must agree not to have any financial interest in their decision outside of the game. If they do not agree, another participant will be randomly selected. No outside bribes or bets in derivative markets. And they must divest their shares in this market - I will purchase the shares at fair market price (i.e. market closing price, unless the price changed substantially near the end, in which case I reserve the right to adjust as needed). If they hold >1000 shares then I reserve the right to randomly select a different participant so that divesting isn't too expensive.
Related
Related questions
🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ124 | |
| 2 | Ṁ39 | |
| 3 | Ṁ34 | |
| 4 | Ṁ26 | |
| 5 | Ṁ26 |
@MartinRandall I guess not, I was imagining it would be cleaner to be able to see the final prediction of the market before the randomized player selection and after. But if people want it reopened, sure.
@jack Oh, people do have limit orders set based on the assumption that it would close and stay closed.
Note I will need some way to contact you to send you the manalink at the end of the game - if you're on the Manifold discord DM me (I'm @jack) or submit a way for me to contact you here: https://forms.gle/RkxRTczUDP9WPBoH8
And to summarize the rules for the player:
The player must sell their shares in this first market at the market price of 78%. (Here that's still profitable for you and you get much more profits from actually playing the game, so that shouldn't be a problem.)
There will be a second market specifically to predict the player's answer. The player is welcome to talk about their thinking, or to say nothing. For all "in-game" discussions, deception is allowed and part of the game.
The player must agree not to have any financial interest in their decision outside of the game. They are not allowed to trade in the market, in derivative markets, or take outside bribes.
@NikitaSkovoroda Thanks, we carried out the divestment of shares and I set up the market for part 2
Ok, random player selection time! There are 38 users in the market, will generate a random number and the numbering will be going top to bottom, first the 20 yes holders on page 1, then the 10 no holders on page 1, and then the 8 yes holders on page 2.
If the user is invalid (e.g. a bot) I'll reroll.
Kind of a pet peeve: to precommit to an action means to set things up so that alternative actions become impossible or costly. So when you say "I precommit to do X", you're making a claim that (as a result of you saying that) if you don't do X, people will react negatively enough and you'll care about their reaction enough to make doing X not worth it. That claim is often not clearly true. It seems to me that when people say "I precommit to do X", it would usually be more accurate (and more understandable to non-nerds) to instead say "I promise to do X".
@StevenK thanks for helping my understanding of this. :) I didn't know that. Would social costs from a broken promise not count as sufficient for a promise to be precommitment? (I could imagine a world where you could promise to anonymously donate, or precommit to publicly donate.)
@NcyRocks I think the cost of publicly breaking a public promise to one box is clearly more than the m1 in box A so the precommitment terminology fits here.
@MartinRandall I agree it's very likely a sufficient precommitment here (though a rare trollish person might value public promise-breaking), it just feels potentially confusing to have a performative utterance also have content about one's utility payoffs.
@JoshuaAnderson I don't know, it's possible that I'm being too peevish and that people in practice understand that "I precommit" means "I hereby take on whatever precommitment costs arise from my having said that" and that it doesn't logically imply that these costs are sufficient to prevent them from later rationally (in terms of their game theory payoffs) taking the action they precommitted against.
@StevenK sure, a trollish anon might make a precommitment statement that is false. And sometimes people make precommitment claims that are outlandish.
I'm trying to imagine an alternate version of a market like this. What do you think of this setup:
Box A contains Ṁ100.
Box B contains Ṁ200 times PROB of market at close.
Select a responder who swears that they have no external financial interest (e.g. side bets) and who has fewer than 100 shares in the market. (OR select such a responder ahead of time and make it known as part of the market, and make different markets for different people, so we can compare how they fare.)
The responder will have their prize adjusted such that it cancels out the profit and loss of any shares they have in the market at close.
Market rules explicitly allow potential responders to lie about how they will respond, and all participants in the market are warned about this in advance.
I want a situation where a two-boxer can make between Ṁ100-300, while a one-boxer can make between Ṁ0-200, but a known two-boxer makes Ṁ100 and a known one-boxer makes Ṁ200.
I also don't want the responder's decision to be affected by their own bets, and I want to limit reputational effects. (I don't think that's entirely possible unless both the responder and the market participants were anonymous to each other, though, which is less fun.)
Do you see any problems with this?
I suppose it's possible that someone could bet NO, then respond YES, but make more money on their NO bet than they won from their response. So it would be impossible to "adjust" their prize to cancel out their market winnings. But I think you could fix this by giving an additional fixed Ṁ100 to the responder no matter what they chose?
@EMcNeill I think it works, the rules you describe are similar to my current rules after the updates I made to require the player to divest and agree not to have any financial interest, which I think is a cleaner way of doing it. Of course your mana amounts are different which I predict will change the strategy significantly.
Yes, it will be explicit in my second market that the responder does not have to be honest in-game. I think the better way to adjust the game to avoid this issue is to make it anonymous, see this discussion thread below
@jack Yeah, I like your updates and I’m reaching for the same thing. I just want to see it with much closer mana amounts. Plus I want to enable a non-anonymous version: I’m tickled by the idea of making multiple markets for different people at the same time, to see who performs best.