What will white (Manifold) play in move 3?
Check the game here: https://lichess.org/GF9YULQP
The game so far: 1. e4 e5 2. Bc4 Nf6
Previous move:
For each response, the average probability in the last four hours before close is measured. With 75% probability, two moves will be randomly drawn, with weight proportional to those market probabilities. With 25% probabilities, three moves will be randomly drawn in the same way. Then for each of the two or three candidate moves, a conditional market is created.
The score of each move will be determined by the average probability in the last 4 hours. The move with higher score will be chosen (and the corresponding condditional market will resolve to the score one move later. The other market(s) will resolve N/A).
More details here:
https://manifold.markets/harfe/will-white-win-in-manifold-plays-ch
The moves "Resign1", "Resign2", "Resign3" are legal moves.
Invalid moves or duplicate moves will be removed from consideration.
🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ5,347 | |
| 2 | Ṁ519 | |
| 3 | Ṁ261 | |
| 4 | Ṁ243 | |
| 5 | Ṁ126 |
@jack well I’m the idiot who totally bailed on one of them, everyone else wins either way so I doubt they’d be willing. Therefore, I’m sure you’ll agree, I’m your only chance of recovering some of it. So I’ll bribe you with ᛗ10 for Nc3. Besides, the other move is just boring, you’ll have way more fun with all the tactical tricks, including the most blunders per move, if you go with my choice.
Payoffs:
Trader d3 Nc3 diff
------ ---- ---- ----
Martin 1903 599 1304
Anton 625 146 479
Andrew 316 122 194
Fion 374 295 79
Daniel 0 279 -279
harfe 3952 5727 -1775
Total 7170 7168 2 (I think due to truncation)@jack If those aren’t convincing enough (as if!), here’s one more that might persuade you: all of us in that list, except one, have relatively modestly-sized balances. The only one with somewhat deep pockets is the one with the most unbalanced betting between the two moves (other than myself), thus I sense he’d be willing to give some betting resistance against your takeover, which of course means you’d recoup more by betting against him.
@deagol great stats. As thanks I will save you the m10 bribe by letting you know that I'm not planning to do any more betting that the limits I already set.
@MartinRandall thanks and you’re welcome. I’m not planning to bet much either way, just a few pennies here and there. And I had a whole spreadsheet to enter trades and compute the avergae prob in the last few hours, but I guess that’s a waste now.
Good luck! (I really don’t care which move wins, just messing with @jack a bit)
The conditional markets are here:
https://manifold.markets/harfe/manifold-plays-chess-3-if-we-play-3 Nc3
https://manifold.markets/harfe/manifold-plays-chess-3-if-we-play-3-ba5e2ae61ecb d3
@harfe any thoughts on the huge incentive you, as creator and organizer, now have in influencing which conditional wins? As well as everyone else with a position here on those moves, but that’s just as intended, except way more intense this time and perhaps unintentionally and arbitrarily distorted.
@deagol now that the markets are created I don't have more influence than a normal bettor, I think?
Ok, I could change the close time while my move is ahead, but that is not something I will do.
Also, I made a bet shortly after market creation, but this should not hugely influence the result.
@harfe Of course, I know you wouldn’t do any of those. What I mean is (and this not at all judging just wondering how you think about it), exclusively about the unusually biased betting behavior (which I agree your condition as creator gives you no advantage) and the profit incentive distorting the underlying premise of the whole structure: “honestly bet on the move that results in a higher score” (as @citrinitas pointed out below). Obviously on this move we all can profit by betting the move which gives the higher payout here, even if incurring a smaller loss on that bet in the winning conditional. In essence we don’t really care that much about what “score for the next move” to aim at.
What I meant by “you as creator and organizer” was, presumably you believe your own design intends to surface the players’ preferences and expert knowledge about the moves, but this market’s accidental yet very fortuitous bounty seems to deny that aspect for all of us and not just manipulators, this particular time alone. Sorry if I’m not explaining myself clearly.
@deagol To put it more bluntly: this bounty will turn us all, including you the very same creator and promoter of the structure, into “dishonest manipulators” (betting just for profit regardless of our beliefs about the best moves).
@deagol I think there should be similar incentive sizes in favor of both moves, plus there may be other bettors who can try to profit from manipulation.
@deagol Ah, I understand now. I think I somewhat agree with you, that now bettors are incentivized to make bets that don't correspond to their true beliefs. We are a bit lucky that both moves are good. Also, this is an unusual situation, which I don't expect to repeat in this way.
I still think that Nc3 would have an advantage against Qh5 even if the stakes are high.
@deagol I was already thinking after move 2 that maybe choosing the move using softmax would be better than argmax.
@harfe for sure it’s better than Qh5 😛 but do you now prefer it over d3 (which I sensed was your preferred between those during the normal market making betting prior to last minute)?
Can you elaborate on softmax vs. argmax? Something about exponential and a vector of probabilities, but can’t quite wrap my mind around how it might apply here.
@deagol example for softmax: if both have the same score, do a coinflip. If difference is 0.01, then pick the higher one with prob = 2/3. If difference is 0.02, then pick the higher one with prob = 4/5. And so on... if difference is 0.0k, then pick higher one with prob = 2^k / (2^k+1).
Yes you are right, my preferences between d3 and Nc3 have suspiciously changed now. Actually, I wanted to make a last-second bet on d3 too, but it was too late and the market already closed.
@harfe sounds cool, so my move 2 “manipulation” to get it at a mere 0.0017 difference would have resolved for my Bc4 with 53% chance, right? I’d have to work much harder, and even near total domination would not guarantee success. It also puts more pressure on the free response to make sure only good moves pass. How does it work with three moves?