Remove cheaters from database

Hey! Would it be possible to remove games played by flagged cheaters from the lichess database?

When studying openings you sometimes come across lines with very high win rates, only to realize that all the games were played by cheaters...

I like the percentages in the database, because unlike an engine, which is completely objective, win rates tell you more about how difficult something is to play for actual human beings. When this is obscured by cheaters it loses a lot of its value.

Hey! Would it be possible to remove games played by flagged cheaters from the lichess database? When studying openings you sometimes come across lines with very high win rates, only to realize that all the games were played by cheaters... I like the percentages in the database, because unlike an engine, which is completely objective, win rates tell you more about how difficult something is to play for actual human beings. When this is obscured by cheaters it loses a lot of its value.

cashcow8

You're only supposed to be looking at these percentages between games when you're doing preparation. Obviously you're not allowed to look at them whilst the game is going on. If you see a line that you want to play, you'll be analysing what the engine thinks too and also looking in the master's database to see if it has been played there.

I have encountered many "inferior" moves that were very popular on lichess but not on the Master's database (usually the one I chose to play that the engine later told me was an inaccuracy).

There are many other factors that can distort win rates: opening traps into which many players have fallen but if avoided lead to a good position for the defending side. Games played at different speeds, e.g. bullet may favour one side whereas classic favours the other... You're only supposed to be looking at these percentages between games when you're doing preparation. Obviously you're not allowed to look at them whilst the game is going on. If you see a line that you want to play, you'll be analysing what the engine thinks too and also looking in the master's database to see if it has been played there. I have encountered many "inferior" moves that were very popular on lichess but not on the Master's database (usually the one I chose to play that the engine later told me was an inaccuracy).

fiskaren

In principle i personally agree, cheater games should not be honored. But there is a problem: The best known method to catch cheaters, is by machine learning. And as of today that needs still a lot of raw data. Hence the games of users flagged as cheaters could be very important for the health of lichess in the future and IMHO in itself should be kept.
It might me possible to filter out unwanted games somehow.

In principle i personally agree, cheater games should not be honored. But there is a problem: The best known method to catch cheaters, is by machine learning. And as of today that needs still a lot of raw data. Hence the games of users flagged as cheaters could be very important for the health of lichess in the future and IMHO in itself should be kept. It might me possible to filter out unwanted games somehow.

InkyDarkBird

edited

@cashcow8 said in #2:

There are many other factors that can distort win rates: opening traps into which many players have fallen but if avoided lead to a good position for the defending side. Games played at different speeds, e.g. bullet may favour one side whereas classic favours the other...
Those factors are all part of normal chess.
What you are trying to argue is that the only chess games placed in the database are the ones where the games are "fair" which is ridiculously restrictive and unnecessary.

@fiskaren said in #3:

In principle i personally agree, cheater games should not be honored. But there is a problem: The best known method to catch cheaters, is by machine learning. And as of today that needs still a lot of raw data.
No they don't.
Just go to all the current or previously flagged cheaters' games and eliminate them from the database.
You also have no evidence as to how much power the cheat detection method consumes.

@cashcow8 said in #2: > There are many other factors that can distort win rates: opening traps into which many players have fallen but if avoided lead to a good position for the defending side. Games played at different speeds, e.g. bullet may favour one side whereas classic favours the other... Those factors are all part of normal chess. What you are trying to argue is that the only chess games placed in the database are the ones where the games are "fair" which is ridiculously restrictive and unnecessary. @fiskaren said in #3: > In principle i personally agree, cheater games should not be honored. But there is a problem: The best known method to catch cheaters, is by machine learning. And as of today that needs still a lot of raw data. No they don't. Just go to all the current or previously flagged cheaters' games and eliminate them from the database. You also have no evidence as to how much power the cheat detection method consumes.

Vxp2023

Today my opponent spent time but I loose? What was that a cheat?

InkyDarkBird

@Vxp2023 said in #5:

Today my opponent spent time but I loose? What was that a cheat?
Do not hijack others' topics. Go make your own topic.

@Vxp2023 said in #5: > Today my opponent spent time but I loose? What was that a cheat? Do not hijack others' topics. Go make your own topic.

fiskaren

@InkyDarkBird said in #4:

... And as of today that needs still a lot of raw data.

No they don't.
Just go to all the current or previously flagged cheaters' games and eliminate them from the database.
You also have no evidence as to how much power the cheat detection method consumes.

Wow, i must have missed a development. Can you explain how it works or link to relevant theoretical papers?
This time please without insulting anyone. Is this possible?

@InkyDarkBird said in #4: ... And as of today that needs still a lot of raw data. > > No they don't. > Just go to all the current or previously flagged cheaters' games and eliminate them from the database. > You also have no evidence as to how much power the cheat detection method consumes. Wow, i must have missed a development. Can you explain how it works or link to relevant theoretical papers? This time please without insulting anyone. Is this possible?

InkyDarkBird

edited

@fiskaren said in #7:

... And as of today that needs still a lot of raw data.
Wow, i must have missed a development. Can you explain how it works or link to relevant theoretical papers?
This time please without insulting anyone. Is this possible?
First off, I did not insult anyone.
Second of all, you were the one making the claim the LICHESS cheating algorithm consumes a ton of data, which you still have provided no evidence for. It is the burden of the affirmative to prove the affirmative, otherwise presumption negates.
In addition, I do not understand how your argument of data consumption would apply to removing games from the game database.

@fiskaren said in #7: > ... And as of today that needs still a lot of raw data. > Wow, i must have missed a development. Can you explain how it works or link to relevant theoretical papers? > This time please without insulting anyone. Is this possible? First off, I did not insult anyone. Second of all, you were the one making the claim the LICHESS cheating algorithm consumes a ton of data, which you still have provided no evidence for. It is the burden of the affirmative to prove the affirmative, otherwise presumption negates. In addition, I do not understand how your argument of data consumption would apply to removing games from the game database.

boilingFrog

Baseline scintillation is the essential issue ... need I say more ?

Vxp2023

#10

@InkyDarkBird said in #6:

Do not hijack others' topics. Go make your own topic.
Sorry, but what do you meaning? I did not understand you completely. You want me not to answer your posts?

@InkyDarkBird said in #6: > Do not hijack others' topics. Go make your own topic. Sorry, but what do you meaning? I did not understand you completely. You want me not to answer your posts?

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