Use of chess engine

Look at this game. On move 12, White makes a serious mistake and loses a knight. At this point, the best thing to do would be to surrender. He doesn't, and from that point on, he never makes a mistake. On move 27, he loses his bishop. But the computer doesn't even consider it a mistake. The same thing happens with move 29, when he loses a rook. This move, according to the computer's analysis, isn't a mistake, but in my opinion it's not a human move. To win, I had to play with 96% accuracy. Black had 79%.
Of course, I'm not sure whether Black used a chess engine, but I have some doubts. Consider that this type of thing happens very frequently.

??? He almost never plays the engine move, and not even T2 or T3 most times.
But ok, first thing I want to say is stop looking at accuracy, cpl and the game analysis in a vacuum. Accuracy is virtually meaningless unless it's considered over many consecutive games and aggregated, and even then can be affected by play style and opening choice. Cpl while more objective, is also affected greatly by how easy the position is to play. From the numbers, I'd suspect you way before the opponent, but you seldom made a move that wasn't also my first choice, even with me going through the game 4x faster than you played it. Move 25 I wanted Rd8, you played Qb3, so I guess that would make me suspicious because I want another rook in the game and so does the engine. :D

As to your interpretation of the analysis and how you see what the analysis flags, once the game is lost, the computer doesn't keep pointing out every small mistake. There won't be any colored text in the notation pointing out either of your mistakes. That doesn't mean the move is approved by the engine. It just means the programmers didn't think saying more about the bad situation would be helpful to anyone. Your opponent made virtually no move reccomend by the engine, and the ever-worsening eval shows as much.

Here's what I do (and this is what I do for suspect games, NOT for my own analysis to learn from, though that is similar.). First and most important, go through the game without the engine. I cannot stress how important this is. See how often the opponent makes a move that totally caught me off guard. Also look for how many mistakes I made that now seem really obvious (yes, that happens). For me, that phase can even be after the computer analysis because I really don't look at that while I'm going through the game. Just don't have the engine on then and certainly don't allow arrows to be drawn while you're looking.

NOW go back and turn on the engine (if you even feel anything stiĺ needs looked at). In this game, I only did because you used this games as an example. With the analysis showing AND the engine on, keep an eye on how often they differ. Yes, the engine which is using more time and cpu, won't always agree with the evals shown in the analysis. They're close but not the same. This game has lots of moves where both of you miss the T1 move, the eval could have been very different. For instance my Rd8 example above, you lost .4 eval for Qb3. Anyway, the computer didn't add notes to the analysis about his allowing the N fork on move 27, but it did change the eval from -7 to -9 because he loses another exchange. You say the computer analysis doesn't call that a mistake but, it just doesn't make any text note of it because by this time the game is so lost, the opponents moves are irrelevant. The criteria for inaccuracy, mistake and blunder are done by percentage and by the late stages of this game, even giving up a bishop doesn't qualify as noteworthy. But don't confuse that with the computer thinking it was a good move.

Honestly, I think if this person had been using an engine, the knight they gave up in the center of the board wouldn't have been nearly enough for you to win. Their defense from move 15 on could have been way better. In fact, here's an experiment for you, go back to the move where you take the bishop with the Rook Pawn, and click on the button for Play from here against the computer with the engine set to maximum and see if you can win. That's what best play looks like. And yes I realize what you're saying is some moves don't make sense, but I can look at your games and your moves and there's plenty of times your moves don't make sense. Everybody sees the moves different, that's why the game is fun.

But in any case, your perception that the end of this game was relatively mistake free is just plain wrong. You really need to look at it with the engine to see just how many suboptimal moves were made, and there were lots of them on both sides. That's why the evaluation graph makes such a slow progress. That's just normal once one side has an advantage. The advantage generally grows slowly over time just like it did here. But it didn't do it here because of perfect play, it did it because of relatively equal imperfect play. In my opinion this opponent has a 0.0% chance they cheated. And I'm not even going to comment about any thoughts of there being engine use and this was actually sandbagging with an engine. That is tin foil hat stuff on an epic scale. This game was just two people playing chess.

@Woland52 said in #51: > Look at this game. On move 12, White makes a serious mistake and loses a knight. At this point, the best thing to do would be to surrender. He doesn't, and from that point on, he never makes a mistake. On move 27, he loses his bishop. But the computer doesn't even consider it a mistake. The same thing happens with move 29, when he loses a rook. This move, according to the computer's analysis, isn't a mistake, but in my opinion it's not a human move. To win, I had to play with 96% accuracy. Black had 79%. > Of course, I'm not sure whether Black used a chess engine, but I have some doubts. Consider that this type of thing happens very frequently. ??? He almost never plays the engine move, and not even T2 or T3 most times. But ok, first thing I want to say is stop looking at accuracy, cpl and the game analysis in a vacuum. Accuracy is virtually meaningless unless it's considered over many consecutive games and aggregated, and even then can be affected by play style and opening choice. Cpl while more objective, is also affected greatly by how easy the position is to play. From the numbers, I'd suspect you way before the opponent, but you seldom made a move that wasn't also my first choice, even with me going through the game 4x faster than you played it. Move 25 I wanted Rd8, you played Qb3, so I guess that would make me suspicious because I want another rook in the game and so does the engine. :D As to your interpretation of the analysis and how you see what the analysis flags, once the game is lost, the computer doesn't keep pointing out every small mistake. There won't be any colored text in the notation pointing out either of your mistakes. That doesn't mean the move is approved by the engine. It just means the programmers didn't think saying more about the bad situation would be helpful to anyone. Your opponent made virtually no move reccomend by the engine, and the ever-worsening eval shows as much. Here's what I do (and this is what I do for suspect games, NOT for my own analysis to learn from, though that is similar.). First and most important, go through the game without the engine. I cannot stress how important this is. See how often the opponent makes a move that totally caught me off guard. Also look for how many mistakes I made that now seem really obvious (yes, that happens). For me, that phase can even be after the computer analysis because I really don't look at that while I'm going through the game. Just don't have the engine on then and certainly don't allow arrows to be drawn while you're looking. NOW go back and turn on the engine (if you even feel anything stiĺ needs looked at). In this game, I only did because you used this games as an example. With the analysis showing AND the engine on, keep an eye on how often they differ. Yes, the engine which is using more time and cpu, won't always agree with the evals shown in the analysis. They're close but not the same. This game has lots of moves where both of you miss the T1 move, the eval could have been very different. For instance my Rd8 example above, you lost .4 eval for Qb3. Anyway, the computer didn't add notes to the analysis about his allowing the N fork on move 27, but it did change the eval from -7 to -9 because he loses another exchange. You say the computer analysis doesn't call that a mistake but, it just doesn't make any text note of it because by this time the game is so lost, the opponents moves are irrelevant. The criteria for inaccuracy, mistake and blunder are done by percentage and by the late stages of this game, even giving up a bishop doesn't qualify as noteworthy. But don't confuse that with the computer thinking it was a good move. Honestly, I think if this person had been using an engine, the knight they gave up in the center of the board wouldn't have been nearly enough for you to win. Their defense from move 15 on could have been way better. In fact, here's an experiment for you, go back to the move where you take the bishop with the Rook Pawn, and click on the button for Play from here against the computer with the engine set to maximum and see if you can win. That's what best play looks like. And yes I realize what you're saying is some moves don't make sense, but I can look at your games and your moves and there's plenty of times your moves don't make sense. Everybody sees the moves different, that's why the game is fun. But in any case, your perception that the end of this game was relatively mistake free is just plain wrong. You really need to look at it with the engine to see just how many suboptimal moves were made, and there were lots of them on both sides. That's why the evaluation graph makes such a slow progress. That's just normal once one side has an advantage. The advantage generally grows slowly over time just like it did here. But it didn't do it here because of perfect play, it did it because of relatively equal imperfect play. In my opinion this opponent has a 0.0% chance they cheated. And I'm not even going to comment about any thoughts of there being engine use and this was actually sandbagging with an engine. That is tin foil hat stuff on an epic scale. This game was just two people playing chess.

LeechessMothsRGhey

#52

@V1g1yy

One can basically "cheat" when they play correspondence, correct?

@V1g1yy One can basically "cheat" when they play correspondence, correct?

V1g1yy

#53

@LeechessMothsRGhey said in #52:

@V1g1yy

One can basically "cheat" when they play correspondence, correct?

I don't understand the context of your question, what are you getting at? We weren't talking about correspondence anyway.

@LeechessMothsRGhey said in #52: > @V1g1yy > > One can basically "cheat" when they play correspondence, correct? I don't understand the context of your question, what are you getting at? We weren't talking about correspondence anyway.

LeechessMothsRGhey

#54

@V1g1yy

What I'm getting at is that your perspectives on detecting cheating during live games seems at odds with your play history. You have 10 games. All are correspondence, games during which you are free to consult external materials.

Seems odd, to say the least. Are you on Lichess just to use the forums?

@V1g1yy What I'm getting at is that your perspectives on detecting cheating during live games seems at odds with your play history. You have 10 games. All are correspondence, games during which you are free to consult external materials. Seems odd, to say the least. Are you on Lichess just to use the forums?

V1g1yy

#55

@LeechessMothsRGhey said in #54:

What I'm getting at is that your perspectives on detecting cheating during live games seems at odds with your play history. You have 10 games. All are correspondence, games during which you are free to consult external materials.

Seems odd, to say the least. Are you on Lichess just to use the forums?
I came to Lichess back when I was receiving coaching and the coach did their lessons here. I played on cc for years, and kept my online games there. In my last year I joined the cheating forum to keep up on the opinions there. It was a ___show officially, but there are people in there who are very capable at writing cheat detection utilities and the results they show are compelling.

I AM one of those people who 'gets a feeling' when playing, but I like to think I'm a little more objective than most at seeing who's using assistance and who's not. In my experience, my 'guessing' has been pretty phenomenal. Now, I can't speak for how many I didn't suspect, but the ones I suspected were nearly 100% caught before it was all over with. That's not hard over there since there's math models and charts showing the % of cheating is overwhelming.

I don't think the fact that I gave up on online chess really disqualifies me from having an opinion in the forum.

As to your question then (correspondence), I still don't understand what you mean. Do you mean you're allowed to use opening books, or that the games take so long that some people do cheat? Or do you mean that the ICCF allows engine use (keep in mind, at least on paper, Lichess and cc do not allow them)?

In my correspondence experience, which was primarily in the USTCL on cc, there are individuals there who do play fair. How many? It depends from team to team. On my team, I didn't see anyone who was tearing up the track, but then our team captain cared enough to toss anyone suspected of anything underhanded. On other teams, there were cheaters and the team captains did nothing about it. 0.00 cpl for 30+ moves is something Magnus doesn't do in classical OTB, but you see that over there and it's not even questioned. Over there 100+ games in a row like that isn't enough evidence.

My point on that is, for the folks who like to bust on the fair play people here on LiChess, these people are 10000% better than the ones over there. I have never seen an egregious case here. Over there, streamers can make videos while they pair one cheater with another and do that for hours on end, no end in sight. It's entertaining, but it's also indicative of how bad it is over there. No such accounts exist here.

@LeechessMothsRGhey said in #54: > What I'm getting at is that your perspectives on detecting cheating during live games seems at odds with your play history. You have 10 games. All are correspondence, games during which you are free to consult external materials. > > Seems odd, to say the least. Are you on Lichess just to use the forums? I came to Lichess back when I was receiving coaching and the coach did their lessons here. I played on cc for years, and kept my online games there. In my last year I joined the cheating forum to keep up on the opinions there. It was a ___show officially, but there are people in there who are very capable at writing cheat detection utilities and the results they show are compelling. I AM one of those people who 'gets a feeling' when playing, but I like to think I'm a little more objective than most at seeing who's using assistance and who's not. In my experience, my 'guessing' has been pretty phenomenal. Now, I can't speak for how many I didn't suspect, but the ones I suspected were nearly 100% caught before it was all over with. That's not hard over there since there's math models and charts showing the % of cheating is overwhelming. I don't think the fact that I gave up on online chess really disqualifies me from having an opinion in the forum. As to your question then (correspondence), I still don't understand what you mean. Do you mean you're allowed to use opening books, or that the games take so long that some people do cheat? Or do you mean that the ICCF allows engine use (keep in mind, at least on paper, Lichess and cc do not allow them)? In my correspondence experience, which was primarily in the USTCL on cc, there are individuals there who do play fair. How many? It depends from team to team. On my team, I didn't see anyone who was tearing up the track, but then our team captain cared enough to toss anyone suspected of anything underhanded. On other teams, there were cheaters and the team captains did nothing about it. 0.00 cpl for 30+ moves is something Magnus doesn't do in classical OTB, but you see that over there and it's not even questioned. Over there 100+ games in a row like that isn't enough evidence. My point on that is, for the folks who like to bust on the fair play people here on LiChess, these people are 10000% better than the ones over there. I have never seen an egregious case here. Over there, streamers can make videos while they pair one cheater with another and do that for hours on end, no end in sight. It's entertaining, but it's also indicative of how bad it is over there. No such accounts exist here.

defeatmoaning

#56

You can't be caught if you are smart cheating. That ls only using engine assistance when you are absolutely not sure what is happening in the position or you don't know what to do, critical moments.

You don't have to use the top engine moves, or just make a bunch of 'bad' moves and then play all accurate moves. Like that guy did on chesscom.

So there is no way to prove it, you are basically asked to write a thesis why they cheated when reporting.

Sometimes their accuracy in games will vary wildly, from 40% to 96%, but obviously 99% of people here believe accuracy has no meaning.

You can't be caught if you are smart cheating. That ls only using engine assistance when you are absolutely not sure what is happening in the position or you don't know what to do, critical moments. You don't have to use the top engine moves, or just make a bunch of 'bad' moves and then play all accurate moves. Like that guy did on chesscom. So there is no way to prove it, you are basically asked to write a thesis why they cheated when reporting. Sometimes their accuracy in games will vary wildly, from 40% to 96%, but obviously 99% of people here believe accuracy has no meaning.

V1g1yy edited

#57

2 you don't have to use the top engine move but most of them do, and if you play enough suboptimal moves your opponent can beat you.

3 I've never written one so I don't know what's involved. I never reported a single player in all my years on the other cesspool site. I figured if I was already paying for their "world class cheat detection", then they certainly didn't need me.

4 in really egregious cases, accuracy is absolutely useful. But as you just pointed out, the very same person can play two games in a row with wildly different accuracy. This isn't me making this up, every person who's ever played two games on this site knows that their own accuracy varies wildly. What are the highest and lowest numbers you've had? Now consider if I try to say that you've played suspiciously because your game has a high accuracy number. Are you seeing how absurd this is? I'm not saying the accuracy is totally meaningless, but what I want to really see is that you found a bunch of moves that stockfish 17 thinks are really good ones and I absolutely couldn't see them if my life depended on it. Now I know something's fishy.

Keep in mind also that I as a player who doesn't cheat, has all sorts of absolutely positively guaranteed good data to compare. And not only that, but it Compares well with my opponents because if I were to play online, those people would be reasonably close in rating to myself. So I should be able to expect they play about like I do. Now I find this to be the case OTB, but I did not find it to be the case on the other site. If you check the video earlier in this thread with Fabi on Levatov, I made reference to a timestamp late in the video where Fabiano says basically you have two choices, you can accept that there will be some cheaters or you can stop playing online chess. But there's your two options.

To address your paragraphs above I'll just call them one two three and four. 1 cases like this are where you're playing a player who is still otherwise pretty good at chess. The vast majority of patzer cheaters are not capable of doing this and they're easier to spot than you might think. I will not say however that it's easy to catch in code, meaning some mathematical computation using an engine. The only engine that works for that is a brain. 2 you don't have to use the top engine move but most of them do, and if you play enough suboptimal moves your opponent can beat you. 3 I've never written one so I don't know what's involved. I never reported a single player in all my years on the other cesspool site. I figured if I was already paying for their "world class cheat detection", then they certainly didn't need me. 4 in really egregious cases, accuracy is absolutely useful. But as you just pointed out, the very same person can play two games in a row with wildly different accuracy. This isn't me making this up, every person who's ever played two games on this site knows that their own accuracy varies wildly. What are the highest and lowest numbers you've had? Now consider if I try to say that you've played suspiciously because your game has a high accuracy number. Are you seeing how absurd this is? I'm not saying the accuracy is totally meaningless, but what I want to really see is that you found a bunch of moves that stockfish 17 thinks are really good ones and I absolutely couldn't see them if my life depended on it. Now I know something's fishy. Keep in mind also that I as a player who doesn't cheat, has all sorts of absolutely positively guaranteed good data to compare. And not only that, but it Compares well with my opponents because if I were to play online, those people would be reasonably close in rating to myself. So I should be able to expect they play about like I do. Now I find this to be the case OTB, but I did not find it to be the case on the other site. If you check the video earlier in this thread with Fabi on Levatov, I made reference to a timestamp late in the video where Fabiano says basically you have two choices, you can accept that there will be some cheaters or you can stop playing online chess. But there's your two options.

This topic has been archived and can no longer be replied to.