[MISRESOLVED BY IAN] How many language models will I beat at chess before I lose to one?
11
1.4kṀ1285
resolved Sep 16
ResolvedN/A
Resolved
N/A
≥15
Resolved
N/A
≥20
Resolved
N/A
≥25
Resolved
N/A
≥30
Resolved
N/A
≥35
Resolved
N/A
≥40
Resolved
N/A
≥45
Resolved
N/A
≥50
Resolved
N/A
≥55
Resolved
N/A
≥60
Resolved
N/A
≥65
Resolved
N/A
≥70
Resolved
N/A
≥75
Resolved
N/A
≥80

Resolves to the number of options on the above market that resolve YES before one resolves NO. Only options naming a specific LLM count.

Currently, that number is thirteen.

Get
Ṁ1,000
to start trading!
Sort by:

Hey, what are you doing? I had positions on this, some I had won already.

@AlanTennant Evan had not beaten 15 models yet, right? The other market that this one references didn't unilaterally resolve N/A

@SpiritofEvanonManifold lol.

Rolling back the site a few hours to fix the fact that you trashed it is hardly a cover-up.

If it's still possible for you to resolve your markets despite being banned, then I could see an argument for not resolving these, but idk about that

@SimonWestlake I'm very bullish on LLM's, but it's fairly impossible to lose against them in good faith provided you know even a little bit about how to play and pay some attention to the game because they fail entirely to play games like chess once you get past openings and common board positions.

I have no steak in whatever else people are arguing about.

@AlanTennant The user was kicked from the site after abusing it for a few hours. Seeing as how he's not a user anymore, nor can he be trusted, I'm not sure how we could resolve these in good faith, hence the N/A. Sorry for the bad experience!

@ian I didn't know this, thank you for your reply, it's really appreciated, and I see, this seems reasonable.

@AlanTennant it looks like he compensated you for your shares, thankfully

@mods this one needs resolving to N/A as well

Depends how good at prompting you are, I bet

bought Ṁ50 YES

Strictly LLM only, or can they perform intermediate steps like writing a chess solver in Python and then running the code and using the answer as part of its answer, like Chat-GPT can?

@AlanTennant No external tools allowed, so it can't run code.

@evan excellent

How would you describe your level of ability at chess?

@AlanTennant Beginner level.

Will you allow them to make illegal moves, if not then will you give them unlimited attempts to make legal moves, or disqualify them, or select a random move?

@AlanTennant Illegal moves will be disallowed and on the third attempt to make an illegal move the model will be to considered to have lost the game.

@evan You've got this one in the bag then, just play protectivly until the game state is atypical, for example almost any game by your 6th turn, then watch as it's rote memorisation of how chess games between experienced players usually start turns into confusion, and then "how does the non-existent queen I ridiculously gave away for free move again, it's on the board right" bewilderment.

© Manifold Markets, Inc.TermsPrivacy