We're playing as black, against lichess's level 3 stockfish bot as white. Will we win?
Resolves YES if we get a checkmate / lichess resigns, or NO if we're mated by lichess, or PROB at 50% if it's a draw, or N/A if I'm unable to continue for some reason.
Link to the game: https://lichess.org/sLmZPb9nTcN
The current move is
I'll post a separate market for each turn, that will close daily at midnight PST. Anyone who holds YES on this market can vote for a move in the comments.
Last time manifold played chess (https://manifold.markets/group/manifold-plays-chess), we lost -- someone bought a bunch of NO and then bid up bad moves. This is an experiment to see if that can be avoided with good market structure.
Could we also include the latest move list in the description? Thanks.
1. g3 d5
2. Bg2 e5
3. Nf3 Bd6
4. c3 Nf6
5. O-O d4
6. Ng5 …
I previously discussed in the comments on Game 1 and on discord ideas for how to structure these markets to work better. I wrote up my ideas here:
@jack what I did was executing moves recommended by stockfish without much risk. What you did was making a move that’s suboptimal, holding more No than Yes shares.
I’m not interested in making markets worse just because it’s allowed to manipulate them.
So please don’t make people note that I did the thing you did first.
@ms_test Fair enough, all I meant was the strategy of using an alt to buy YES shares risk-free. As for the manipulation, according to the author the point of this market was to test whether it could be manipulated (as described in the last line of the market description).
I would have given the same warning to @ms_test if he'd started pursuing his own inscrutable goals, but as he says his votes were just stockfish.
Manipulation is entirely in accord with the letter and spirit of the rules. I'm treating this as a low-stakes futarchy lab, and the precise way the rules fail is noteworthy.
Shouldn’t this market be temporarily closed at midnight every day during resolution of the move markets? Then reopen unless move is “resign” or stockfish replies with checkmate or resign. I think this would prevent someone buying a ton of YES here and voting “resign” just before midnight, and immediately swinging to NO here the moment the move is known. I did just that last night (with my puny capital) to sneak in a few votes (except I voted for a decent move).
@deagol The intended defense against that is I might take a bit before evaluating the votes. If you buy a bunch of YES and !VOTE Resign, there's a long and non-deterministic period where someone can notice that and tank your YES investment before you have a chance to sell.
So far I've been pretty quick to run the evaluation, but you should not assume that will definitely continue.
@citrinitas So, @jack now holds 130k YES hedged by their alt @jack2, about 90% of the votes at near zero risk (I think). And his current vote is a sucky move, so they could vote “resign” last minute, have a bot read the fairly random reply, then immediately unwind here if evaluation picks it. I think his position was like 2k NO before recent takeover here. But who knows he might as well change his vote to a decent move and let us keep playing.
@deagol Sure, but the only thing he'd win in that "unwind" is limit orders + the subsidy. Right now there's a grand total of about M$429 total that could be earned by moving the market to 0%, mostly from the AMM subsidy, so I don't think anyone in particular would be losing out there.
Yeah, there are a small amount of profits available for someone doing this but I don't think it's a big deal currently and doesn't seem worth the hassle of constantly having to adjust the close date. It could become a bigger deal if people put up big limit orders though.
The only way this works is if there are conditional markets for every move. This doesn't work, sorry
Of course not, but someone else can easily do it too
@jack has managed to accumulate a position of 821 NO on @jfjurchen's meta market, and is currently holding 98 YES here. There are 496 YES shares voting in the current policy market, so that's a 16% chance @jack can snipe with "Resign", unless something else changes.
How will this play out? Let's see!
Intended exploit here is:
1. I buy NO on this market down to 15%
2. I create the meta-market and take a YES position, which hedges things and also makes it profitable for traders to buy YES on the meta-market and NO here
3. Someone else buys YES on this market, hedges with even more NO on the meta-market (which is an arbitrage), and votes for blunders
4. Manifold loses chess
5. Hopefully I profit and so does someone else
I'm not certain it'll work, but this still has the "YES bettors need to win the vote every time, NO only needs to win once" property, so I'm hopeful.
@jfjurchen The two markets will be arbitraged, and if they are arbitraged well enough then you won't be able to build up a NO stake there and a YES stake here.
In step 3, that only works if this market is lower than the other market. I expect it to go the other way - this market is like voting shares, and the other market is non-voting shares, and voting shares should trade at a premium.
Oh nvm, you are saying that you are deliberately moving the prices so that it's profitable for someone else to arb them, and then to exploit the markets. I think the problem with that is by definition if you arb the markets, you have no incentive either way.
Yeah, that's definitely a possibility. But I'd want to take a net NO position in any case (because I'm optimistic that some exploit will work) and hedging on a second market that I get new trader bonuses on seems high-upside.
Why only level 3?
I expect the dominant factor will be how many blunders we're able to survive, not how well we play on average. I'm guessing that level 3 will be very beatable if the market remains well-aligned, but will be able to beat us if someone figures out a way to force bad moves.
Btw link to the previous game: https://manifold.markets/citrinitas/will-white-win-in-alexliesmans-mani