Originally posted by lauseyYes, also this would pick up on people legitimately analysing finished games with an engine. I've analysed a few of the banned players games in the past to see if I agreed with the verdict, and it's pretty clear in the cases I checked that one or both players were using engines.
A minor point as you revealed it wouldn't be a viable option anyway, but cookies cannot detect what applications you are running.
A cookie file is just pure data which websites can read/create/modify. Unless software applications are specifically programmed to create or modify certain cookies (which I very much doubt Fritz or any other chess engine will do ...[text shortened]... ne's PC to install executable code (which can possibly detect what applications are running).
Originally posted by DeepThoughtThere could be a way that it will work using cookies, but not be practical. Just mentioning this for argument sake. 😉
Yes, also this would pick up on people legitimately analysing finished games with an engine. I've analysed a few of the banned players games in the past to see if I agreed with the verdict, and it's pretty clear in the cases I checked that one or both players were using engines.
Fritz and all other chess engines could be programmed to modify an existing cookie and place in the FEN of the move made on there. The RHP website can then "ping" this same cookie after each move on the site and see if the FEN matches up. 😀
Of course, this will require ALL chess engines to be programmed to modify cookies to place in FENs for the plan to be effective. 🙂
EDIT: It will also be too easy for the user to circumvent. All he/she has to do is delete the cookie after each move in the engine.
Originally posted by DeepThoughtI had a recent K&Q vs K&R ending that I eventually won after about 25 moves due to a blunder by my opponent. My book on end games by Averback clearly had this as a win and I was studying this furiously and slowly getting there. To win required creating Zugswang repeatedly by transposing from a position in which white had the move to an identical position in which black had the move.
Match up rates in themselves don't really tell you all that much, I'm pretty convinced most people on this site will agree with Fritz or Crafty or whatever on what move to play next in this position:
[fen]4k3/7R/4K3/8/8/8/8/8[/fen]
I assume that the games mods have some way of assessing how easy a move is to find - where the example above counts as ...[text shortened]... effort into the games, and its annoying that my work was wasted on an imbecile with an engine.
After the game I put this on Fritz which assessed me as +5 pawns and then stumbled around mindlessly for 50 moves before drawing. Any strong player would know this is a win for the stronger side and would know how to do it. An engine would fail and a tablebase would be too precise (even that strong player would probably make the odd inaccurate move). Similarly I had K, P and 2Ns vs K which Fritz continued to think was a win long after the pawn was moved into the drawing zone. Again this is elementary as any book on Knight endings will show how to do this and where the pawn should be, so a strong player would not play on in such circumstances although again most would need a tablebase to actually win it.
This is a shame. I chatted with him a bit (we're both .Net developers) but only once about chess. I thought he had winning chances (but it was probably drawn) in a game and he just handed me the draw. When I asked about his passed c-pawn (after accespting) he just said "it wouldn't make much difference".
Originally posted by Dragon FireEngines are pants at minimal material endgames, basically the tree pruning algorithm relies on finding moves that are so poor it can quickly work out that it doesn't need to calculate their consequences. In the case of K+Q v K+ R almost every position evaluates to the same score of +4 (depending on what value the engine gives to the queen and rook) as a consequence of which the engine has to search every line of play including patently ridiculous ones and if the checkmate is further away than whatever its search depth is it won't find it.
I had a recent K&Q vs K&R ending that I eventually won after about 25 moves due to a blunder by my opponent. My book on end games by Averback clearly had this as a win and I was studying this furiously and slowly getting there. To win required creating Zugswang repeatedly by transposing from a position in which white had the move to an identical position ...[text shortened]... not play on in such circumstances although again most would need a tablebase to actually win it.
According to EGTB the longest checkmate with K+Q v K+R is 35 moves, an engine won't get much beyond a 20 move search even with that limited amount of material as the branching factor is more or less the same as the number of legal moves. So it's not a surprise that Fritz couldn't find the checkmate.
Originally posted by DeepThoughtUnless the engine has access to tablebases (as long as there is a tablebase for the pieces that is left on the board), then it wouldn't even need to do any calculation.
Engines are pants at minimal material endgames, basically the tree pruning algorithm relies on finding moves that are so poor it can quickly work out that it doesn't need to calculate their consequences. In the case of K+Q v K+ R almost every position evaluates to the same score of +4 (depending on what value the engine gives to the queen and rook) as a ...[text shortened]... the number of legal moves. So it's not a surprise that Fritz couldn't find the checkmate.
Originally posted by Sicilian SmaugIn order to play at ICC you have to use their computer program called Blitzln, which can tell when you switch from their program to another program. A friend of mine who plays at this site got "caught" by ICC's program cause he was doing just that - had Fritz open while playing in a game at the same time.
This wouldn't be effective ( I can't comment on if it is even possible as I'm no computer expert).
We are permitted to analyse finished games in Fritz or other engines. If a player has Fritz on in the back ground doing a full analysis on a finished game whilst in another window they are continuing with their RHP games then such a system would penalise them.
My friend says he was not using Fritz to cheat, but simply had it open from looking at another game, but they gave him a "C" behind his name anyways. Yeah, it doesn't sound right, but on the other hand, why was he switching back and forth between the two programs as he played? I mean, you can't sit down to play someone in a tournament and open up a chess book.
Originally posted by lauseySo there are 2 ways here to conclusively find an engine abuser.
Unless the engine has access to tablebases (as long as there is a tablebase for the pieces that is left on the board), then it wouldn't even need to do any calculation.
1. In a "minimum material endgame" that is a "book win" the computer just stumbles around and draws; or
2. If the computer does not stumble then it plays a move perfect 35 move forced win (from a tablebase).
In either case it becomes obvious that a human is not playing.
I have had (at least) 2 games that fall into this category so it will occur regularly with a long term user but it may take 300+ games for it to become obvious.
Originally posted by Dragon FireWe probably are already deducing what methods the mods are using to catch engine abusers. 😉
So there are 2 ways here to conclusively find an engine abuser.
1. In a "minimum material endgame" that is a "book win" the computer just stumbles around and draws; or
2. If the computer does not stumble then it plays a move perfect 35 move forced win (from a tablebase).
In either case it becomes obvious that a human is not playing.
I have had ...[text shortened]... l occur regularly with a long term user but it may take 300+ games for it to become obvious.
Chessbase boot cheaters regularly, one of the players that I lost to was booted shortly after, heres the game.
[Event "Rated game, 90m + 0s"]
[Site "Tournaments"]
[Date "2006.09.02"]
[Round "?"]
[White "Caffeinated"]
[Black "HenKathi"]
[Result "0-1"]
[ECO "B14"]
[WhiteElo "1834"]
[BlackElo "1976"]
[PlyCount "78"]
[EventDate "2006.09.17"]
[TimeControl "5400"]
1. e4 {9} c6 {5} 2. d4 {16} d5 {3} 3. exd5 {16} cxd5 {10} 4. c4 {9} Nf6 {21} 5.
Nc3 {5} e6 {38} 6. Nf3 {9} Bb4 {31} 7. cxd5 {33} Nxd5 {26} 8. Bd2 {23} Nc6 {106
} 9. Bd3 {279} O-O {107} 10. O-O {16} Be7 {439} 11. Qe2 {16} Bd7 {114} 12. Nxd5
{36} exd5 {10} 13. a3 {472} Bf6 {95} 14. Be3 {159} Bg4 {138} 15. h3 {247} Bh5 {
28} 16. Rac1 {724} Qd7 {235} 17. Rfd1 {685} Rae8 {199} 18. Bb1 {62} Bxd4 {248}
19. Bxh7+ {726} Kxh7 {14} 20. Ng5+ {77} Kg6 {114} 21. Qd3+ {207} f5 {225} 22.
g4 {429} Bxe3 {8} 23. fxe3 {93} Kxg5 {126} 24. gxh5 {27} Kxh5 {2} 25. Rc5 {114}
Qc7 {180} 26. Kf2 {181} Qh2+ {14} 27. Ke1 {20} Qxh3 {52} 28. Kd2 {27} Ne5 {7}
29. Qe2+ {17} Ng4 {36} 30. Rc3 {125} d4 {23} 31. Rd3 {49} dxe3+ {19} 32. Kc1 {7
} Rc8+ {40} 33. Kb1 {6} f4 {47} 34. Rd5+ {64} Kh6 {7} 35. Rg1 {26} f3 {14} 36.
Qd3 {69} f2 {32} 37. Rd6+ {33} Rf6 {2} 38. Rxf6+ {63} gxf6 {8} 39. Rxg4 {11}
f1=Q+ {Caffeinated resigns (Lag: Av=0.39s, max=3.9s) 11} 0-1
While playing the game one move struck me as odd and even now looking at it after the move seems total unhuman.
Move 25...
Ask yourself what you'd play in this position for black....then check the move played in the game......its very odd, I cant work out why the engines seem to like the game move more than the human move......although maybe I cant work it out because its outside my understanding 🙂
Originally posted by Sicilian SmaugIndeed! Qd6 is the move just about every human says.....the engine/game move is very odd from a human perspective 🙂
Think I'd play Qd6 to threaten both the rook and Qg3+
Given the chance to hit the rook with tempi and threaten lots of nasty things or putting the queen oppsite a rook no matter how undeadly it might look......hmm hard choice 😉