I've been facing a relatively large number of players lately who on going through the computer analysis are making no or nearly no mistakes and having centipawn loss < 20. I don't want to be paranoid or waste the staff's time investigating, so what's your criteria? Oddly enough having played hundreds of games this has only become an issue extremely recently - like within the past couple of weeks.
I've been facing a relatively large number of players lately who on going through the computer analysis are making no or nearly no mistakes and having centipawn loss < 20. I don't want to be paranoid or waste the staff's time investigating, so what's your criteria? Oddly enough having played hundreds of games this has only become an issue extremely recently - like within the past couple of weeks.
Anyone who cheats on lichess.org will eventually be caught by the lichess software. If you read some of the previous articles in http://en.lichess.org/blog Thibault shows you how he included an algorithm that checks if a player is cheating.
Here is an article titled, "How To Catch A Chess Cheater: Ken Regan Finds Moves Out Of Mind" which an be found here: http://www.uschess.org/content/view/12677/763
It should be noted that short games are much more likely to show a very small centipawn loss--if any. A 20 centipawn loss is not uncommon in slower time controls. However, if you are playing lots of bullet time controls and most of your opponents are averaging less than a 5 centipawn loss per game and they have not already been caught by the software, then I would be highly suspicious about their games and I would consider reporting them to the admins.
Anyone who cheats on lichess.org will eventually be caught by the lichess software. If you read some of the previous articles in http://en.lichess.org/blog Thibault shows you how he included an algorithm that checks if a player is cheating.
Here is an article titled, "How To Catch A Chess Cheater: Ken Regan Finds Moves Out Of Mind" which an be found here: http://www.uschess.org/content/view/12677/763
It should be noted that short games are much more likely to show a very small centipawn loss--if any. A 20 centipawn loss is not uncommon in slower time controls. However, if you are playing lots of bullet time controls and most of your opponents are averaging less than a 5 centipawn loss per game and they have not already been caught by the software, then I would be highly suspicious about their games and I would consider reporting them to the admins.
The system is automatically finding the obvious cases and marks them, and when finding suspicious one they get auto-reported for the mods to look at.
However, there is a few cases when it doesn't catch them early. You can just add a comment in the report saying something like "unsure if cheating, pls do a quick check". Low centipawn loss is more common for games with a few number of moves or longer time controls.
You can then check their profile some day later and see if they got marked, if so then make more reports. But otherwise your intuition might be wrong and you should probably make fewer reports.
If there is no comment and they aren't suspicious, then I usually spend more time trying to figure out why they was reported and especially going though recent games with the reporter.
The system is automatically finding the obvious cases and marks them, and when finding suspicious one they get auto-reported for the mods to look at.
However, there is a few cases when it doesn't catch them early. You can just add a comment in the report saying something like "unsure if cheating, pls do a quick check". Low centipawn loss is more common for games with a few number of moves or longer time controls.
You can then check their profile some day later and see if they got marked, if so then make more reports. But otherwise your intuition might be wrong and you should probably make fewer reports.
If there is no comment and they aren't suspicious, then I usually spend more time trying to figure out why they was reported and especially going though recent games with the reporter.
Report early as long as you can convince yourself that the moderator reviewing it won't regret seeing such a report. Usually that means including evidence (the suspected cheating games) and showing your investigation efforts, if any.
Generally from my own experience, you shouldn't need to worry about taking up the staff's time since there's always a backlog in reports anyway. Mass reporting only becomes a problem when most of previous reports are known to be dubious or outright misled.
Report early as long as you can convince yourself that the moderator reviewing it won't regret seeing such a report. Usually that means including evidence (the suspected cheating games) and showing your investigation efforts, if any.
Generally from my own experience, you shouldn't need to worry about taking up the staff's time since there's always a backlog in reports anyway. Mass reporting only becomes a problem when most of previous reports are known to be dubious or outright misled.
The really blatant ones get caught quickly. But I'm more talking about borderline cases where it's a mix of a decent human player using computer assistance as opposed to inputting every single stockfish top move. Those guys are a lot trickier. One of the most recent ones I suspected played more than 1,000 games before getting busted.
I wonder if running the computer analysis on their games is helpful to the automated systems? Basically I don't want to waste people's time with cases I'm not 100% on, and these type of players it's almost impossible to be 100% on.
The really blatant ones get caught quickly. But I'm more talking about borderline cases where it's a mix of a decent human player using computer assistance as opposed to inputting every single stockfish top move. Those guys are a lot trickier. One of the most recent ones I suspected played more than 1,000 games before getting busted.
I wonder if running the computer analysis on their games is helpful to the automated systems? Basically I don't want to waste people's time with cases I'm not 100% on, and these type of players it's almost impossible to be 100% on.
Tricky cheaters are still cheaters.
Tricky cheaters are still cheaters.
Yes, run analysis on all games you are suspicious about. They are used by the auto-reporter and helps mods not waste time waiting for them to complete.
Mods have a completely different overview for players games, they can more easily see if a player is suspicious (so reporting even when uncertain is good). But even with our overview, adding links to the games in the comment might be helpful, especially for correspondence games and some of the variants that don't have computer analyse yet. Don't overdo it though, we rarely need more than 5-10 games from the reporter.
For these difficult ones, if you are suspicious about certain moves or behavior it's helpful if you include that as a comment.
Yes, run analysis on all games you are suspicious about. They are used by the auto-reporter and helps mods not waste time waiting for them to complete.
Mods have a completely different overview for players games, they can more easily see if a player is suspicious (so reporting even when uncertain is good). But even with our overview, adding links to the games in the comment might be helpful, especially for correspondence games and some of the variants that don't have computer analyse yet. Don't overdo it though, we rarely need more than 5-10 games from the reporter.
For these difficult ones, if you are suspicious about certain moves or behavior it's helpful if you include that as a comment.
Okay, cool. I'll be more active with the analyzer and shoot a report over occasionally on the more like than not borderline guys.
Okay, cool. I'll be more active with the analyzer and shoot a report over occasionally on the more like than not borderline guys.
When in doubt report!
I find the centipawn loss confusing, there are loads of games I have seen where one side has made no mistakes but the eval ends up at +1/2. That doesnt make sense to me, +1 should be winning! I know it would probably be quite difficult but if there was an option to get the top 3 lines from Stockfish that would be great!
When in doubt report!
I find the centipawn loss confusing, there are loads of games I have seen where one side has made no mistakes but the eval ends up at +1/2. That doesnt make sense to me, +1 should be winning! I know it would probably be quite difficult but if there was an option to get the top 3 lines from Stockfish that would be great!