Release : Chess System Tal EXTREME

[mclane]

The funny thing ist IMO the first thing the masses do is let CSTal or absurd play against stockfish.
As if this makes any sense at all. But they do it.
And then - because it is not winning, they post "how weak it is".

Oh man.

[Leto]

I'm having issues running Absurd and Chess System Tal Extreme on the Aquarium GUI, they refuse to run on that GUI. I am able to run them fine on the Nimzo 3D GUI.

[Rebel]

If all sacrificial attacks worked they wouldn’t be sacrificial attacks. If they always worked Cstal would be 200 Elo above SF in the rating lists, and it isn’t, so they don’t.
What exactly do you want?

I was not really asking for anything here, just sharing my thoughts and observations. I think that CSTal/Rebel is a most interesting experiment to improve on the perceived beauty of the games and I follow its development with interest.

To also answer to the two following posts - if I understood correctly that there is some kind of "parameter" "extremism" that you can vary in training: I wonder how CSTal games with this leading to an ELO 2500 and ELO 2200 level would look like. A 3000 ELO level if real is still just too strong to make it viable as an opponent for humans.

Absurd 1.0 (or better Absurd-3000 as it turned out) is (was) just an example how easy it now is to create a higher ranked style engine once you have created a solid network controlled by a parameter. It takes one day to create the network and one day to test it, how convenient. To make an elo rated Absurd-2500 now is a piece of cake, however I am not going to put my brand name to it, neither can I stand the garbage moves it surely will play. Absurd-3000 is already over the top.

You think 2500 Elo is garbage..?

Making too many false sacrifices.

Code: Select all

 STYLE   King          Short           
 Total  Attack   %     Games   %     Sacs    %    Engine
706305  228566 24.1%  164006 51.6%  313733 56.6%  Absurd 1.0  
524401  214038 23.2%  153444 57.7%  156919 38.5%  Chess System Tal Extreme  
511122  213311 23.7%  149143 57.4%  148668 37.4%  Rebel Extreme 1.1  
486486  165661 25.4%  171709 60.5%  149116 35.8%  Patricia 3.01  
482444  185940 25.3%  161609 61.3%  134895 36.1%  CSTal 2.1 EAS  
407780  158700 27.0%   99659 49.7%  149421 40.8%  Patricia 5 xtremeEAS  
397519  135669 23.4%  164415 64.3%   97435 26.7%  Rebel EAS 3.0 

Look at the sacrifices, 313.733, almost double than Chess System Tal Extreme, nice, but also a sing of many false sacrifices, over the top definitely.

By engine standards, yes, but for human preparation it's not at all. To anyone on this forum I'd say the same thing: try over the board to defend against an aggressive FIDE 2500 Elo player, it is a daunting task.

Playing Chess System Tal Extreme or Rebel Extreme 1.1 will drive any GM crazy IMO.

My guess is 2800 - 3100 Elo is the sweet spot so I suppose you've achieved the goal. However, this type of thing is undervalued by the community IMO as club players to IM's are looking for interesting ideas all the time in their prep.

Indeed, but then use Chess System Tal Extreme, Rebel Extreme 1.1 and Patricia 5.0 for good ideas, not Absurd 1.0

Nevertheless I will give it a try to proof my point, why not.

[Peter Berger]

+1

I am not so convinced about your +1 here, given that you can't stand the garbage moves and wouldn't want to stick your brand name to it if taken to more of an extreme.
Anyway, Chris and you have made your concept clear.

To something else: at the slow time controls I enjoy to observe there seems to be a problem with the EBF of Absurd 1.0. The low depths work fast enough per usual but from about depth 20 it often just gets stuck somewhere. I didn't notice this phenomenon with CS Tal Extreme. Currently I don't get the impression that it might perform at 3000 level at slower time controls against say Crafty as it simply can't outsearch it as all other versions usually do. Is this plausible by concept - or would it be useful to collect data on this?

[chrisw]

+1

I am not so convinced about your +1 here, given that you can't stand the garbage moves and wouldn't want to stick your brand name to it if taken to more of an extreme.
Anyway, Chris and you have made your concept clear.

To something else: at the slow time controls I enjoy to observe there seems to be a problem with the EBF of Absurd 1.0. The low depths work fast enough per usual but from about depth 20 it often just gets stuck somewhere. I didn't notice this phenomenon with CS Tal Extreme. Currently I don't get the impression that it might perform at 3000 level at slower time controls against say Crafty as it simply can't outsearch it as all other versions usually do. Is this plausible by concept - or would it be useful to collect data on this?

If it generates wild deep lines with lots of checks (highly plausible with Absurd) then there going to be a lot less prying and a lot more extension of the wild positions. Cstal has plenty of extension and “do more search” heuristics for these cases. May cause EBF++. What are you complaining about, we said it gets weaker with more crazy?
I know, we’ll make it crazy without mistakes and get 400 Elo over SF. Or we’ll weaken it so weak players stand a chance, but then you’ll complain it plays losing moves. Or we’ll make it exciting and sacrificial and you say sometimes it loses. I give up.

[chrisw]

I'm having issues running Absurd and Chess System Tal Extreme on the Aquarium GUI, they refuse to run on that GUI. I am able to run them fine on the Nimzo 3D GUI.

I’ll correct the typos above. GUIs that try to be too clever by half often can’t run bog standard chess engines that conform to protocol.

[Rebel]

+1

I am not so convinced about your +1 here, given that you can't stand the garbage moves and wouldn't want to stick your brand name to it if taken to more of an extreme.
Anyway, Chris and you have made your concept clear.

To something else: at the slow time controls I enjoy to observe there seems to be a problem with the EBF of Absurd 1.0. The low depths work fast enough per usual but from about depth 20 it often just gets stuck somewhere. I didn't notice this phenomenon with CS Tal Extreme. Currently I don't get the impression that it might perform at 3000 level at slower time controls against say Crafty as it simply can't outsearch it as all other versions usually do. Is this plausible by concept - or would it be useful to collect data on this?

Chris already answered you, hope you get what he said.

Like to add, the game against Crafty proves the repeated points I am making, no plan, many garbage moves. That won't happen with CSTAL-Extreme and Rebel-Extreme 1.1, there is always something to play for.

Consider Absurd 1.0 as entertainment.

Replay at the amazing games it can (emphasis added) play, especially : the Games of ABSURD 1.0 sorted on sacrifices

https://rebel7775.wixsite.com/rebel/absurd-1-0

[Leto]

I'm having issues running Absurd and Chess System Tal Extreme on the Aquarium GUI, they refuse to run on that GUI. I am able to run them fine on the Nimzo 3D GUI.

I’ll correct the typos above. GUIs that try to be too clever by half often can’t run bog standard chess engines that conform to protocol.

Yes I see your point this does seem to be more of a ChessOk Aquarium GUI issue as I'm also able to run those two engines fine on Fritz 20 GUI. I'll contact their support, I'll be surprised if I get a response and moreso if they manage to fix it.

[Leto]

A game between Chess System Tal Extreme and chess.com's Magnus Carlsen bot which I think plays around the 2800-3000 computer chess level. Out of curiosity I forced CSTE to play the accelerated London but everything after 2.Bf4 is CSTE:

[pgn]
[Event "Play vs Bot"]
[Site "Chess.com"]
[Date "2026.02.14"]
[Round "?"]
[White "Chess System Tal Extreme"]
[Black "Magnus"]
[Result "1-0"]
[BlackElo "2882"]
[WhiteElo "?"]
[Termination "by checkmate"]
[ECO "A45"]
[EndDate "2026.02.14"]
[Link "https://www.chess.com/game/computer/729293137"]

1. d4 Nf6 {I'm the 16th undisputed world chess champion...} 2. Bf4 c5 {...and
I've been the number #1 player in the world since 2011...} 3. d5 d6 {...let's
see how long you last. Good luck $1} 4. Nc3 {Smart move.} 4... e5 {I don't really
make small talk when I play. I can't control what the bot version of me does,
though.} 5. Bg5 Be7 {This is a fun way of developing.} 6. e4 {That seems like a
smart, reasonable move.} 6... O-O {King safety is important indeed $1} 7. Bd2 Bd7
{The real me would probably play a better move, but the bot me will settle for
that.} 8. a4 Na6 {The tension is building.} 9. Bb5 {My rating is almost 2900. Is
there any tougher bot you can find $2} 9... Bxb5 {I think I'll take.} 10. axb5 {My
knight is starting to feel some pressure.} 10... Nc7 {I can appreciate a good
knight every once in awhile.} 11. Nge2 {Okay.} 11... Qd7 {I think I had this
position in a game against my father when I was 8. I won.} 12. Ng3 {This is an
interesting position.} 12... g6 {I've won the world title in every time
control.} 13. Bh6 Rfc8 {You are stronger than I thought. Is that true $2 I don't
know. I was programmed to say that.} 14. O-O {It’s nice of you to take care of
your king $1} 14... Nxb5 {Takes.} 15. Nxb5 {Very reasonable, yes.} 15... Qxb5
{Let's slide this piece over here. Don't worry, it's a good move.} 16. Ra3 {I
like that move for you.} 16... Rc7 {Hmm.} 17. Nf5 {I would stand and applaud
that move if I wasn't a bot.} 17... Ne8 {I started to play chess because my
sisters were better than me and I wanted to win. Now $2 I think I'm slightly
better.} 18. Qg4 Qxb2 {I don't think the best move is obvious here.} 19. Bc1
{It's good to see you find defensive moves like that.} 19... Qxc2 {Let's see
what you do about this pawn.} 20. Nh6+ {I like that move for you.} 20... Kg7 {I
normally do what my intuition tells me to. Most of the time spent thinking is
just to double-check.} 21. Rf3 {That is a fantastic find.} 21... Bf6 {I'm quite
good in positions with a strong bishop like this. Also, all other positions.}
22. h4 Qa4 {I really appreciate how powerful the queen is. Such a fun piece to
move.} 23. h5 Kf8 {Tweet at me if you beat my bot so I can ignore it.} 24. hxg6
{That move is not bad.} 24... hxg6 {I can always get out of a jam like that.}
25. Bg5 {Self-confidence is very important. It is better to overestimate your
prospects than underestimate them.} 25... Bxg5 {Seems like a reasonable trade.}
26. Qxg5 Qxe4 {Even I need to defend sometime.} 27. Ng4 Qd4 {My dad, Henrik, was
rated almost 2100. He's probably better than your dad.} 28. Qxg6 {You're down
material, but I quite like your position.} 28... e4 {I don't like to pat myself
on the back, but I will for that move.} 29. Rf4 {You're making some good moves.}
29... c4 {This move is excellent, if I do say so myself.} 30. Qh7 Qg7 31. Qh2
{Most people take my trades. Interesting choice.} 31... c3 {I hope my bot
doesn't say something every time I make the best move. It's probably quite
annoying.} 32. Ne3 {I really like that move. Pat yourself on the back.} 32... c2
33. Nf5 Qf6 34. Rc1 Rc5 35. Qh7 Rxd5 {Let's take.} 36. Rxc2 Rd8 {I don't want to
calculate this too deep, let's just play a move.} 37. Rc1 Re5 {If you're
wondering, that was the best move.} 38. g3 Rxf5 39. Rxf5 Qe6 40. Re1 {Good
move.} 40... f6 41. Rh5 d5 42. Rc1 a5 {This will do.} 43. Qg6 Qg8 {Let's go
here.} 44. Qf5 e3 45. Rh7 exf2+ {You've got some thinking to do now.} 46. Kg2 {I
respect you for challenging me. But I don't fear you.} 46... Rd6 47. Rc8 d4 48.
Kf1 Qxh7 49. Qxh7 {Not many people can reach an endgame against me. And even
less win.} 49... Re6 {Pieces are happier when they can see more squares.} 50.
Rb8 Re1+ {I was one of Time Magazine's 100 most influential people when I was
22. I will be 82 by the time this game finishes.} 51. Kxf2 {Not many people have
lasted 50 moves in a game against me. You have, of course, because this is a bot
version of me. You get the idea.} 51... Re7 52. Qg6 a4 53. Kg2 Rf7 54. g4 {Sure,
why not.} 54... a3 55. g5 fxg5 56. Qd6+ {How dare you $2} 56... Re7 {Let's step
out of that.} 57. Qf6+ {It seems like you've created a few threats...} 57... Rf7
58. Qh8+ Ke7 {Let's sidestep here.} 59. Qe5+ {When you play me, your king is
never safe.} 59... Kd7 60. Qd5+ Ke7 61. Rxb7+ Nc7 {Let's block that.} 62. Rxc7+
Kf6 63. Rxf7+ Kg6 64. Qe6+ Kh5 65. Rh7# {Wow, you got me $1 How about a rematch $2}
1-0
[/pgn]