Ideas for dealing with engines.

Originally posted by forevergreenwithenvy
A. you will be pleased to hear this guy is easy to catch, he's not very good at chess so he lets his engine do all the work, plays engine moves from start to finish, he never questions the engines moves, never shows it altern ...[text shortened]... he hand. he might not run all the time, unless he gets hooked.

So if anyone has 20 games that match totally up to an engine, say fritz8, is pretty much cheating right ?

Originally posted by no1marauder
...I don't think it's "singling" weaker players out; it is just making the commonsense observation that anyone with any experience in chess knows: players DO NOT make the type of amazing improvement in their games that these two ...[text shortened]... e principles you would expect even an average club player to know.

I think contrast to past playing experience is a very interesting area, and acknowledge it may be a start, if not attempting to solve the global problem, but at least part of the problem. However, there are some important issues that need to be highlighted even with this approach of selective investigation :

(1) The first one is that juniors can improve quite dramatically in their first two/three years of chess. If they did start their first year with very poor games here, their 2nd/3rd year might be significantly improved. There might be some greater correlation with *any* engine as their tactical ability improves. Especially if like a lot of juniors they are in fact also being coached by an International Master or a Grandmaster. Added to this is (2)

(2) Openings - The possibility of matching an Engine in the opening phase of the game is made possible and legal because anyone has the right to look up the opening in an openings database. It could be in certain forcing lines of certain openings such as the Marshall gambit, that the first 25 moves match any engine - because Marshall gambit theory is very deep!

(3) Past track record in the real world - Players with very little track record of a strong rating in the wider world of chess. Apart from David Tebb, and Ironmaster (who is an IM?!), who are the other players in the top 10? I think I have beaten David Tebb myself in the British Major open a few years ago, and appreciate he is a strong Bcf 190ish player who has even beaten Kasparov in a simultaneous. But who are these other unknown entities? By the very same argument of a lack of previous track record in the real world of chess, where do these 2000+ players suddenly emerge who can play amazing tactics? What were their real-world ratings before? Are they immune to investigation because they have played a strong game from day 1 here?

(4) What is a 90%+ Engine match? An engine can be set thinking on a postition from the period of time of say 30 seconds to 10 days. At one point is the Engine move obtained for matching purposes with a player? Or is it all the candidate moves the engine thought were reasonably good?

Originally posted by Tryfon Gavriel
I think contrast to past playing experience is a very interesting area, and acknowledge it may be a start, if not attempting to solve the global problem, but at least part of the problem. However, there are some important issues that need to be highlighted even with this approach of selective investigation :

(1) The first one is that juniors can i ...[text shortened]... urposes with a player? Or is it all the candidate moves the engine thought were reasonably good?

In response:


1) Juniors CAN improve dramatically with time; Juniors (or anybody else) CAN'T go from playing barely adequate chess against other inexperienced players to defeating very good players on a regular basis in a matter of weeks. Plus there was no gradual improvement in these two players' games even though they were constanly playing: on Day X they were legitimate 1200 players, on Day X + 1, they were legitimate 2000+ players. This is impossible to do without cheating;

2) Note that Grayeye's analyses generally omit the opening for the reasons you've given. Using opening databases PROPERLY (and you have to have some advanced chess knowledge to do even that) can avoid crushing defeats early; it can't get you a bunch of middlegame, tactical wins. At best, opening databases get you out of the opening with an advantage or equality, you still have to win the game in the middle or end games. I don't believe using even the best opening database can raise any player more than a 100 points or so;

3) I'm not interested in a "witch hunt" of every player on this site; I'm only interested in stopping blatant and obvious cheating. I agree that it is difficult to tell most of the time if the player is very strong or using an engine and I usually don't get too worked up about it; as Feivel says you still have to play the board. OTB play is different from play here; most players OTB ratings are roughly comparable but not all - the ability to use books and databases, the relaxed time constraints, the psychological advantage of playing on your home turf, etc. make RHP easier to play well on, to me, than OTB tournaments. I think everybody should be "immune to investigation" unless there's very good reason to believe they're using an engine;

4) Grayeyes can answer that better than I and I think it's a good question.

If you want to call identifying obvious cheats "selective investigation" that's up to you; I prefer to call it "identifying obvious cheats". I would tend to think that any extremely rapid rise of a player's rating who has been on the site for some time would at least raise eyebrows, if the player has some reasonable explanations as to why he's playing better and/or engine analysis of his games was inconclusive that would be good enough for me. I would give the benefit of the doubt to the player and require something like what we call in law "clear and convincing" evidence of cheating before any punishment is imposed. That being said, I believe that standard has absolutely been met concerning Tlai1992 and James Woodley.

If any move has dropped outside the top 2 I havent posted it in analysis. If you are on par with 190 bcf then look at JW games and see for yourself.

The moves the engine pick dont change a lot, they just change order of which the engine thinks is best, makes it easy to use the same engine to look at their games.

I posted some human play games against engine moves (Tournaments thread called James Woodley), you can see the difference in match up is striking.

Originally posted by Grayeyesofsorrow
If any move has dropped outside the top 2 I havent posted it in analysis. If you are on par with 190 bcf then look at JW games and see for yourself.

The moves the engine pick dont change a lot, they just change order of whic ...[text shortened]... es Woodley), you can see the difference in match up is striking.

If you have put this into Fritz 8 and the analysis is shown say in less than two minutes on each move, then this implies to me that engines have improved dramatically, and have become much easier to detect now because they are so fast and strong. Their detection would also therefore be much more practical as well (Less than 30 minutes a game to do a match up?!).

The games seem very well played tactically. For what period of time was the engine allowed to "think" on each move to obtain the top two candidate moves? If this is a secret, please could you message me privately.

You have to appreciate that a proper ongoing correspondence match with ICCF GM Nickel and some Engines has the Engines thinking for hours and days on moves, not just say 30 seconds per position. But perhaps the situation with engine improvement makes detection a lot faster and more practical nowadays?! Perhaps it will become much easier to detect engine users because they (engines) are simply getting so strong? I am seriously asking this, as the only engine I use is for looking at master games and my horrendous otb games sometimes, with Chessbase light, and the engine is a free one - Fritz version 4 I think. It's variations seem quite poor in general except for some blunt tactics it sometimes finds.

Recently I thought I was playing some very advanced version of Fritz on uchess server in a 5 minute game. I posted the 5 minute game a few weeks ago onto the Chess forum, because the game looked like perfect moves from Black. The player is over 2100 now on Uchess which in a way is a good acid-test for engine users because of the very very short reflection times compared to correspondence chess style. If a player has only seconds to think up moves, it is unlikely they can see 15 moves ahead combinations. The game went:-

[Event "uChess rated"]
[Site "www.uchess.com"]
[Date "2004.10.2"]
[Round "?"]
[White "kingscrusher"]
[Black "Lets keep him anonymous"]
[Result "*"]

1. c2-c4 Ng8-f6 2. Nb1-c3 e7-e6 3. g2-g3 c7-c5 4. Bf1-g2 Nb8-c6 5. e2-e4 Bf8-e7 6. d2-d3 d7-d5 7. c4xd5 e6xd5 8. e4xd5 Nc6-b4 9. Ng1-e2 O-O 10. O-O Nb4xd5 11. a2-a3 Rf8-e8 12. Ne2-f4 Nd5xf4 13. Bc1xf4 Qd8-b6 14. Ra1-b1 Bc8-f5 15. Nc3-e4 c5-c4 16. Ne4xf6 Be7xf6 17. d3xc4 Ra8-d8 18. Qd1-f3 Bf5-e4 19. Qf3-g4 Be4xb1 20. Rf1xb1 Bf6-d4 21. Qg4-f3 g7-g5 22. Bf4xg5 Qb6-g6 *

Please can you tell me the % match up of black moves in the above game.

Best wishes
Tryfon

Hello Tryfon

Have you got any background info on this guy, does he play OTB, and if so has he got a rating, and also would it be possible to post a few more of his games.

FGWE

Originally posted by forevergreenwithenvy
Hello Tryfon

Have you got any background info on this guy, does he play OTB, and if so has he got a rating, and also would it be possible to post a few more of his games.

FGWE

I have messaged Russ to ask if possibly in a future version of Uchess that the pgn games can be automatically stored if they are not already. It would be interesting to play through a big sample of games of the two players who are both 2100+.

I did take the opportunity a few weeks after the loss, to question both of them (the two players >2100) when many players were around, in the real-time chat public area. They failed to answer many of the basic questions about chess that i asked. They seem to me to have no chess "culture" or evidence for any playing ability whatsoever.

I believe it is a discredit to uchess that these two players are occuping the first two spots, when really blitz chess is taken seriously by over-the-board players. Many IM's and GM's on the Internet play blitz chess, and uchess could in my view improve with some PGN game analysis and a computer-detection committee. For the simple fact that the reflection time is in seconds for games there, and so perfectly played games are unlikely to be consistently played out unless the players are top-level GMs.

For me the game with black was played close to perfection with the temporary pawn sacrifice creating a weakness of the isolated pawn on d3, which proved to be a losing weakness the way it was later played by Black. The temporary pawn sacrifice is a book-line which was shown to me a few months ago by a 190+ BCF player. That is not what concerns me - it is the pin-point accuracy of every single move, which leaves the impression that one has been "clinically crushed", which is rather unpleasant!

All but one move came out to be the top fritz move, the move that wasnt top fritz was 3rd choice for the engine.

1.c4 Nf6 2.Nc3 e6 3.g3 c5 4.Bg2 Nc6 5.e4 Be7 6.d3 d5 7.cxd5 exd5 8.exd5 Nb4 9.Nge2 0-0 10.0-0 Nbxd5

11.a3 Re8 [Fritz 8: 11...Be6 12.Nf4 Nxf4 13.Bxf4 Qd7 14.Qa4 Qxa4 15.Nxa4 Nd5 16.Be5 f6 17.Nc3 -0.44/10 ]

12.Nf4 Nxf4 [Fritz 8: 12...Nxf4 13.Bxf4 Qb6 14.Rb1 Bf5 15.Qf3 Rad8 16.Bc7 Qxc7 17.Qxf5 Qd7 18.Qxd7 Rxd7 19.Rfe1 Kf8 -0.41/7 ]

13.Bxf4 Qb6 [Fritz 8: 13...Qb6 14.Rb1 Bf5 15.Qf3 Rad8 16.Bc7 Qxc7 17.Qxf5 Qd7 18.Qxd7 Rxd7 19.Rfe1 Kf8 -0.41/8 ]

14.Rb1 Bf5 [Fritz 8: 14...Bf5 15.Qf3 Rad8 16.Bc7 Qxc7 17.Qxf5 Qd7 18.Qxd7 Rxd7 19.Rfe1 Kf8 20.h3 -0.41/11 ]

15.Ne4 c4 [Fritz 8: 15...c4 16.Be3 Qe6 17.Re1 Rad8 18.Nc5 Bxc5 19.Bxc5 Qxe1+ 20.Qxe1 Rxe1+ 21.Rxe1 -0.66/9 ]

16.Nxf6+ Bxf6 [Fritz 8: 16...Bxf6 17.dxc4 Bxb1 18.Qxb1 Be5 19.Be3 Qb3 20.Qe4 Re7 21.Qh4 Bf6 22.Bg5 Qxb2 23.Bxf6 -0.91/6 ]

17.dxc4 Rad8 [Fritz 8: 17...Rad8 18.Qf3 Bxb1 19.Rxb1 Qc5 20.b3 Bd4 21.Rf1 Qxa3 22.Qxb7 Qc5 -0.81/0 ]

18.Qf3 Be4 [Fritz 8: 18...Bxb1 19.Rxb1 Qc5 20.b3 Bd4 21.Rf1 Qxa3 22.Qxb7 Qc5 -0.81/10 ]

19.Qg4 Bxb1 [Fritz 8: 19...Bxb1 20.Rxb1 Bd4 21.Bd5 Bxf2+ 22.Kh1 Rxd5 23.cxd5 Qb3 24.Qh3 Qxd5+ 25.Qg2 Qd3 26.Ra1 Qe4 27.Qxe4 -1.16/0 ]

20.Rxb1 Bd4 [Fritz 8: 20...Bd4 21.Bd5 Bxf2+ 22.Kh1 Rxd5 23.cxd5 Qb3 24.Qh3 Qxd5+ 25.Qg2 Qd3 26.Ra1 Qe4 27.Qxe4 -1.16/10 ]

21.Qf3 g5 [Fritz 8: 21...g5 22.Qg4 Bxf2+ 23.Kh1 h6 24.Rf1 Qxb2 25.Bc7 Rc8 26.Qd7 Qe2 -1.59/6 ]

22.Bxg5 Qg6 [Fritz 8: 22...Qg6 23.Rd1 Qxg5 24.b4 Re6 25.Kh1 b6 26.h3 Red6 -3.16/9 ]

I too have noticed a distinct lack of chess culture amoung some top players. Even things as basic as having no idea how a club game works. Or when you are beat by a player 2000+ and you say "I should have never of left the knight en prise" and they are like "en prise?"
I could see people not getting "retrograde analysis" but en prise.....everyone knows that right?

Originally posted by Grayeyesofsorrow
I too have noticed a distinct lack of chess culture amoung some top players. Even things as basic as having no idea how a club game works. Or when you are beat by a player 2000+ and you say "I should have never of left the knight en prise" and they are like "en prise?"
I could see people not getting "retrograde analysis" but en prise.....everyone knows that right?

Maybe we should translate it 🙂

Feivel

Originally posted by Grayeyesofsorrow
All but one move came out to be the top fritz move, the move that wasnt top fritz was 3rd choice for the engine.

1.c4 Nf6 2.Nc3 e6 3.g3 c5 4.Bg2 Nc6 5.e4 Be7 6.d3 d5 7.cxd5 exd5 8.exd5 Nb4 9.Nge2 0-0 10.0-0 Nbxd5

11.a3 Re8 [Fritz 8: 11...Be6 12.Nf4 Nxf4 13.Bxf4 Qd7 14.Qa4 Qxa4 15.Nxa4 Nd5 16.Be5 f6 17.Nc3 -0.44/10 ]

12.Nf4 Nxf4 [Fritz 8 ...[text shortened]... 9/6 ]

22.Bxg5 Qg6 [Fritz 8: 22...Qg6 23.Rd1 Qxg5 24.b4 Re6 25.Kh1 b6 26.h3 Red6 -3.16/9 ]

For the second time i find myself in agreement with Grayeyes, this is becoming a bad habbit, lol,

The game is an allmost perfect Blitz T/C match with fritz, There's a big eval drop at move 18, fritz wants 18...Bxb1, he plays Be4, but as Grayeyes states Be4 is 3rd choice. How many people would find/play 21...g5 in a 5 min Blitz ?, not shure about that.

In my view black is an engine, fritz strikes again :-)

Originally posted by Tryfon Gavriel

Recently I thought I was playing some very advanced version of Fritz on uchess server in a 5 minute game. I posted the 5 minute game a few weeks ago onto the Chess forum, because the game looked like perfect moves from Black. The p ...[text shortened]... nlikely they can see 15 moves ahead combinations. The game went:-

Have a read through the uchess forum and you'll see a load of posts with the top 2 players being accused of cheating by using engines AND multiple accounts.

D

Apologies for not contributing to thread. Being a a simple man, I can only focus on one problem at a time, and over the last two weeks my mind has been elsewhere. There is some good stuff in this thread too, the classification of the different type of cheats being very interesting.

-Russ

Originally posted by Tryfon Gavriel
I have messaged Russ to ask if possibly in a future version of Uchess that the pgn games can be automatically stored if they are not already. It would be interesting to play through a big sample of games of the two players who are both 2100+.

I did take the opportunity a few weeks after the loss, to question both of them (the two players >2100) wh ...[text shortened]... leaves the impression that one has been "clinically crushed", which is rather unpleasant!

I've beaten the top guy on uchess twice in a 5min game with increments I don't recall. Havent looked to see who is #2 probably beat him too. Anyway lack of book chess knowledge doesn't really mean much though. It sometimes seems that because someone is good at the game people assume they must read every single book and learn this and that variation or this or that opening or endgame. Oh and I am not in anyway defending those 2 people because if someone did go through #1s games they would find something. The one advantage to playing an engine user on uchess if its a short game is that they will eat their time up switching back and forth to the engine. If you can stay in the game you can either run them out of time completely or leave them so short on time they start playing for themselves and screw the game up. Both times i beat the #1 guy he ran almost completely out of time before making some quick bad moves and then resigning before completely out of time. You know what I would like to see on that site though, how about the ability to play more than one timed game at a time. Would love to see the top guy play two 5 min games at one or heck even two 10 min games. I like to play alot of games so something like that would be really nice, taking on like 3 games at once and seeing how i do. Anyway sorry for going off topic this is redhotpawn not uchess they have their own forum.

Originally posted by Grayeyesofsorrow
I too have noticed a distinct lack of chess culture amoung some top players. Even things as basic as having no idea how a club game works. Or when you are beat by a player 2000+ and you say "I should have never of left the knight en prise" and they are like "en prise?"
I could see people not getting "retrograde analysis" but en prise.....everyone knows that right?

ok now not knowing stuff like that and being a 2000+ player I could see questioning that players true ability. Though i'd still like to believe there has to be a couple who really are that good and still don't know.