'Mittens ChessBot' (Chess.com) Stuns Chessworld

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
towforce
Posts: 12700
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by towforce »

Quick question to anyone: I'm guessing nobody has written a tool for testing chess.com bots?

I imagine in would be a bit of an undertaking to set that up: you'd need an automated web application test tool, and you'd need to write a program it to relay moves between the bot and whatever program you were testing against.
Human chess is partly about tactics and strategy, but mostly about memory
User avatar
towforce
Posts: 12700
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by towforce »

carldaman wrote: Wed Jan 11, 2023 9:29 am
Jjaw wrote: Wed Jan 11, 2023 1:13 am If you have questions about mittens chess engine , download it & play against it ... I did ... https://workupload.com/file/gCAw6Eps5nd
If Mittens can be freely downloaded, then maybe someone can run the similarity tool on it to see which engine it resembles the most.

Also, if that download link is real, it ought to be possible to look at the bundle and see whether it includes an NN weights file.

However, in the absence of any evidence that Mittens is stronger than the "non personality" bot at level 25, it's probably, as Larry has speculated, just that (or something similar) with some parameter changes. If so, then yes - what's the fuss about? Anyone watch any of the early videos about Mittens and know why chess experts got excited about it? I certainly got the impression that chess experts thought it was a jump in strength.

There's also the ongoing mystery as to why chess.com turned the strength down after the hype began - which would be easily explained if it was caused by chess.com's cloud servers being overloaded. However, if chess.com's bots do their move selection in the browser, then how would one explain their decision to lower Mittens' strength?
Human chess is partly about tactics and strategy, but mostly about memory
User avatar
towforce
Posts: 12700
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by towforce »

towforce wrote: Thu Jan 12, 2023 12:23 am
carldaman wrote: Wed Jan 11, 2023 9:29 am
Jjaw wrote: Wed Jan 11, 2023 1:13 am If you have questions about mittens chess engine , download it & play against it ... I did ... https://workupload.com/file/gCAw6Eps5nd
If Mittens can be freely downloaded, then maybe someone can run the similarity tool on it to see which engine it resembles the most.

Also, if that download link is real, it ought to be possible to look at the bundle and see whether it includes an NN weights file.

However, in the absence of any evidence that Mittens is stronger than the "non personality" bot at level 25, it's probably, as Larry has speculated, just that (or something similar) with some parameter changes. If so, then yes - what's the fuss about? Anyone watch any of the early videos about Mittens and know why chess experts got excited about it? I certainly got the impression that chess experts thought it was a jump in strength.

There's also the ongoing mystery as to why chess.com turned the strength down after the hype began - which would be easily explained if it was caused by chess.com's cloud servers being overloaded. However, if chess.com's bots do their move selection in the browser, then how would one explain their decision to lower Mittens' strength?

I've just had a look at the 7z (7Zip compressed) file at that link: it contains nothing but a windows executable file.

My guess - somebody has:

* downloaded the WASM executable for Mittens

* there is a decompiler available for WASM: it decompiles to a meaningless C program (because it won't contain the program symbols, unlike many decompiled Windows executables, where you actually get back code that is close to the original)

* recompiled this C program into a Windows executable
Human chess is partly about tactics and strategy, but mostly about memory
supersharp77
Posts: 1266
Joined: Sat Jul 05, 2014 7:54 am
Location: Southwest USA

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by supersharp77 »

M ANSARI wrote: Wed Jan 11, 2023 7:43 am I don't get the big deal about this Mittens engine. I mean the cat pic looks cute, but how is this engine any different than the other hundreds of other engines that are not up to latest SF strength? Is there a specific reason why this engine is making headlines? It just seems like a weaker engine ... which of course would still do really well against any human!
CornfedForever wrote: Wed Jan 11, 2023 8:23 pm Someone asked 'what the big deal' is with Mittens. I think...basically that it talks trash as is beats you. It looked that way in the...think it was a Hikaru video I saw.
And...cats are cool.
Well The Whole Concept Was Absolutely Brilliant...Actually when Gotham Chess Put Out His Video Complaining About Losing to "Mittens" I thought It was a Real Cat!! LMAO....The Whole Idea is Hilarious!! Hahahaha...Lol!! :lol: :wink:

https://esports.gg/news/gaming/what-is- ... s-cat-bot/

https://www.reddit.com/r/chess/comments ... y_a_thing/

Image

Image

Image
User avatar
M ANSARI
Posts: 3734
Joined: Thu Mar 16, 2006 7:10 pm

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by M ANSARI »

I have to admit that Chess.com did get a lot of buzz and whoever decided on this was pretty smart. That is a cute kitty.
lkaufman
Posts: 6279
Joined: Sun Jan 10, 2010 6:15 am
Location: Maryland USA
Full name: Larry Kaufman

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by lkaufman »

towforce wrote: Thu Jan 12, 2023 12:23 am
carldaman wrote: Wed Jan 11, 2023 9:29 am
Jjaw wrote: Wed Jan 11, 2023 1:13 am If you have questions about mittens chess engine , download it & play against it ... I did ... https://workupload.com/file/gCAw6Eps5nd
If Mittens can be freely downloaded, then maybe someone can run the similarity tool on it to see which engine it resembles the most.

Also, if that download link is real, it ought to be possible to look at the bundle and see whether it includes an NN weights file.

However, in the absence of any evidence that Mittens is stronger than the "non personality" bot at level 25, it's probably, as Larry has speculated, just that (or something similar) with some parameter changes. If so, then yes - what's the fuss about? Anyone watch any of the early videos about Mittens and know why chess experts got excited about it? I certainly got the impression that chess experts thought it was a jump in strength.

There's also the ongoing mystery as to why chess.com turned the strength down after the hype began - which would be easily explained if it was caused by chess.com's cloud servers being overloaded. However, if chess.com's bots do their move selection in the browser, then how would one explain their decision to lower Mittens' strength?
The chess experts thought (correctly) that it was a jump in strength from all other named bots, but I don't think they compared it with level 25. Nakamura lost by an overwhelming margin (maybe 15 to 1 or so) against a weaker (older) version of level 25, and I don't get the impression that Mittens would score higher than this against him, it did lose with just b7 pawn odds to him. The lowering of strength was probably to bring it within human range; based on the above match for level 25 no human has more than a tiny chance to score, and perhaps this was deemed undesirable. I think that the strength of level 25 (and perhaps also Mittens) is about 3300 CCRL in blitz or 3200 in Rapid, whereas humans would have lower CCRL ratings than their FIDE ratings, at least in Rapid and especially blitz.
Komodo rules!
Werewolf
Posts: 2058
Joined: Thu Sep 18, 2008 10:24 pm

Re: Re:Alpha Zero 16 Challenges Mittens The cat!!

Post by Werewolf »

AndrewGrant wrote: Mon Jan 09, 2023 1:57 am
Almost certainly a youtuber running a random engine, calling it AlphaZero, and farming content. Which seems to work for him.
I fell for this trick :oops:
Plutie
Posts: 20
Joined: Sun Jan 30, 2022 6:14 am
Full name: Evan Engler

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by Plutie »

towforce wrote: Thu Jan 12, 2023 12:30 am
towforce wrote: Thu Jan 12, 2023 12:23 am
carldaman wrote: Wed Jan 11, 2023 9:29 am
Jjaw wrote: Wed Jan 11, 2023 1:13 am If you have questions about mittens chess engine , download it & play against it ... I did ... https://workupload.com/file/gCAw6Eps5nd
If Mittens can be freely downloaded, then maybe someone can run the similarity tool on it to see which engine it resembles the most.

Also, if that download link is real, it ought to be possible to look at the bundle and see whether it includes an NN weights file.

However, in the absence of any evidence that Mittens is stronger than the "non personality" bot at level 25, it's probably, as Larry has speculated, just that (or something similar) with some parameter changes. If so, then yes - what's the fuss about? Anyone watch any of the early videos about Mittens and know why chess experts got excited about it? I certainly got the impression that chess experts thought it was a jump in strength.

There's also the ongoing mystery as to why chess.com turned the strength down after the hype began - which would be easily explained if it was caused by chess.com's cloud servers being overloaded. However, if chess.com's bots do their move selection in the browser, then how would one explain their decision to lower Mittens' strength?

I've just had a look at the 7z (7Zip compressed) file at that link: it contains nothing but a windows executable file.

My guess - somebody has:

* downloaded the WASM executable for Mittens

* there is a decompiler available for WASM: it decompiles to a meaningless C program (because it won't contain the program symbols, unlike many decompiled Windows executables, where you actually get back code that is close to the original)

* recompiled this C program into a Windows executable
That's an unrelated engine that just happens to also be named "Mittens" - download link found on that CEDR blogspot site, a few pages back. seems to have no relation whatsoever to the bot, which is using some version of Komodo. There's also something going on at CCC right now, which seems to back up what others have said about it, including the settings used.
Fritz 0
Posts: 151
Joined: Fri Mar 11, 2022 12:10 pm
Full name: Branislav Đošić

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by Fritz 0 »

Yesterday I run a 6 game match between Mittens bot and Dragon 3 with Elo set to 3000. I made moves for Dragon, playing against Mittens on chess.com. Dragon won 3,5:2,5 (2 wins, 1 loss, 3 draws). Of course, 6 games is nothing, but I have an impression that Dragon at this level is really somewhat stronger than this version of Mittens. So, I assume we can say that this Mittens bot is certainly below 3000 Elo in Rapid (15+10), or 3300 Elo in Blitz (3+2), in human terms.
Werewolf
Posts: 2058
Joined: Thu Sep 18, 2008 10:24 pm

Re: 'Mittens ChessBot' (Chess.com) Stuns Chessworld

Post by Werewolf »

Did we establish if Mittens is running on a server or the browser powered by the local PC?