Originally posted by doccotoanbaiI'm not sure about Fritz, but lower rated personalities in Chessmaster make asinine blunders to try and make it look like it is weaker. The program will play like an idiot then suddenly turn into a grandmaster at various parts of the game. Not very realistic and easy to spot.
Even every games.
If his opponent rate is 1500, he can adjust engine level to 1700 or dont take the best suggestion. Winning but not dominate his opponent, no beautiful sac, who could think he cheat?
When low rated human players (like me) make weak moves, there is some logic to them. Can't say the same about CM personalities.
Originally posted by UndeadNightOrchey.
[b]I'm not sure about Fritz, but lower rated personalities in Chessmaster make asinine blunders to try and make it look like it is weaker. The program will play like an idiot then suddenly turn into a grandmaster at various parts of the game. Not very realistic and easy to spot.
b]
Thats how many of my tournament opponents play.
Fancy that!
Quick reminder guys - no matter what your suspicions may be, even if you're 100% sure, please do not accuse users of cheating unless they have been officially banned for the offence. I've had to remove 2 such accusations from this thread alone this morning.
If you have suspisions please share them, and any proof, with the game moderation team and not post them in the forums.
Originally posted by UndeadNightOrcI don't know how Chessmaster simulates weak moves, but there are a number of things you could do. The most obvious is just to restrict the search ply depth. You could alter the evaluation function to make it prefer features of positions even in inappropriate cases, for example knights to bishops in both closed and open positions; or add a small signed random number to the result of each call to the evaluation function to make the algorithm fuzzy. To make it blunder you could add random tree pruning (so you occasionally get beta fails when there shouldn't be). The thing is that these methods would just make the engine play pathologically, rather than like a human.
I'm not sure about Fritz, but lower rated personalities in Chessmaster make asinine blunders to try and make it look like it is weaker. The program will play like an idiot then suddenly turn into a grandmaster at various parts of the game. Not very realistic and easy to spot.
When low rated human players (like me) make weak moves, there is some logic to them. Can't say the same about CM personalities.
Humans make mistakes in a very different way. There's emotional content so we over or underestimate our positions, or just plain get lazy when we think we've won. We tend to make pattern recognition errors, so that a positional feature or tactic is missed in all variations, not just the one an engine tree pruning algorithm randomly left out. I think it would be quite hard to make an engine play like an average human.
Originally posted by DeepThoughtThat's the problem simulating human play, we use pattern recognition, whether low rated or high. Since all the progs use more or less brute force, it is very hard if not impossible for a prog to effectively simulate ANY human which is why idiots like Ironman and Jimster get caught.
I don't know how Chessmaster simulates weak moves, but there are a number of things you could do. The most obvious is just to restrict the search ply depth. You could alter the evaluation function to make it prefer features of positions even in inappropriate cases, for example knights to bishops in both closed and open positions; or add a small signed ...[text shortened]... omly left out. I think it would be quite hard to make an engine play like an average human.
Originally posted by wormwoodI was referring more to the fact that some ppl manage to stay around for nearly 3 years, such as Jimster, and yet here there are ppl who arent surprised at him being banned.
why is that surprising? we are required to shut up about our suspicions until the players in question have been banned.
GauraV was a similar case. Seemed like there was a bunch of ppl saying "he was obvious, knew it was coming eventually" but the guy had played 3000+ games at that point!
Whereas Myrthir or MaryAnn, for instance, get ppl snickering in threads just short of saying they are cheaters.
so it is surprising in that so many ppl always seem to know who the cheats are, but they still manage to remain on site for 2+ years.
So how many games have to be alerted to the mods? doesnt seem to be any real consistency in that. I guess some users are more obvious like a Touragnew or MaryAnn. But I never heard a snicker about Jimster, so that it was more surprising to me when he was banned.
Originally posted by DeepThoughtI'm writing my own engine and have been thinking about how to simulate human-like play. I came to the conclusion that you should just randomly make the computer unable to see say 5% of the possible moves on the board. So on a given turn the computer might be unable to see any moves from e4 to g6. The computer, like a human, will thus miss some tactics.
Humans make mistakes in a very different way. There's emotional content so we over or underestimate our positions, or just plain get lazy when we think we've won. We tend to make pattern recognition errors, so that a positional feature or tactic is missed in all variations, not just the one an engine tree pruning algorithm randomly left out. I think it would be quite hard to make an engine play like an average human.
To prevent the computer from doing utterly stupid things like leaving its queen en prise, you could modify the probabilities based on the area of the board, so that the computer pays more attention to some parts of the board than others: moves attacking queens or involving the most recently moved piece will be more likely to be seen, while maybe moves to or from the edges of the board will be more likely to be missed. I haven't tried it out, but I suspect you could get a pretty convincing simulation of lower-level human play if you combined this with reducing search depths and dumbing down the evaluation function.
Originally posted by Forum ModeratorCool login. 😕
Quick reminder guys - no matter what your suspicions may be, even if you're 100% sure, please do [b]not accuse users of cheating unless they have been officially banned for the offence. I've had to remove 2 such accusations from this thread alone this morning.
If you have suspisions please share them, and any proof, with the game moderation team and not post them in the forums.[/b]
In advance, to the stupied peoples, I know it's not a login...thanks.
Originally posted by Ramiri15Black should have resigned about move 50. Any strong player would so engine use again obvious.
Very funny 😀
Game 2591385 Here's another good one. A battle between two engine users ends in the bigger idiot playing until mate 😵.
I found another one a while ago as I was looking through the graphs on the banned players list; it was a game that lasted something like 60 moves that was obviously drawn 20 moves before. But the two players' e ...[text shortened]... Unfortunately I couldn't find it again (though digging for it uncovered that gem up there).
Originally posted by cmsMasterDitto! I am well capable of making the most insane blunders, usually because I am not taking my opponent seriously enough (thinking I have an easy win), I am playing too many games too quickly (like now with this damn tournament that just gave me 18 games) or have had a few drinks too many.
I think I'm one of those retarded engines - I make good moves (Decent anyway...) for a while..then in the middle of it all I'll make a horrible blunder. 🙁 I wish I could stop that...
If (when) that happens I stop moving, study the position carefully and take a little time so naturally I suddenly do start playing better and if my opponent is 400 points lower rated than me I have a fighting chance of clawing my way back.