The antichess and atomic communities here on lichess have a serious problem with players artificially inflating their ratings using bots. This is not a new problem in the variants community. Here is a note from the developer of the atomic bot Sordid (from FICS) that had to stop playing rated games many years ago for this reason.
"I am now playing unrated atomic only. This way I don't have to police for people abusing the rating system. You can repeat openings all you want. Hopefully I will be a better contribution to the atomic community this way."
In atomic there are openings in which bots cannot evaluate correctly. This is because they have an inability to evaluate some endgame style positions correctly which, combined with the horizon effect, creates a weakness. It means there exists positions where bots will almost always transition into losing endgames where many humans would not. This is what players have been abusing.
There are many players that will play exclusively or primarily as white against these bots. They will, sometimes in hundreds of games consecutively, repeat the same lines from the opening. The bots almost always reply with the same inferior move resulting in a loss or draw at best. The player rinses and repeats often to the tune of hundreds of ratings points. In over 10 years of playing variants, the most extreme cases of ratings abuse I've ever seen has recently occurred with @GodelEscherBot. It has recently lost over 600 antichess rating points and 200 atomic ratings points. The antichess abuse is so bad that players have been repeating the same winning GAMES IN THEIR ENTIRETY (to completion!) multiple times against it. At that point, you're not even playing chess.
I think the only solution to this problem is to clear these artificial ratings and prevent bots from playing rated games. I don't think forcing colour alternation is a way of solving this problem. A 1900 player that repeats an exploitable line against a 2500 bot can go 5-5 in a series with relative ease. This would almost certainly not be the case against 2500 rated human opposition.
edit: speling and grammar.
"I am now playing unrated atomic only. This way I don't have to police for people abusing the rating system. You can repeat openings all you want. Hopefully I will be a better contribution to the atomic community this way."
In atomic there are openings in which bots cannot evaluate correctly. This is because they have an inability to evaluate some endgame style positions correctly which, combined with the horizon effect, creates a weakness. It means there exists positions where bots will almost always transition into losing endgames where many humans would not. This is what players have been abusing.
There are many players that will play exclusively or primarily as white against these bots. They will, sometimes in hundreds of games consecutively, repeat the same lines from the opening. The bots almost always reply with the same inferior move resulting in a loss or draw at best. The player rinses and repeats often to the tune of hundreds of ratings points. In over 10 years of playing variants, the most extreme cases of ratings abuse I've ever seen has recently occurred with @GodelEscherBot. It has recently lost over 600 antichess rating points and 200 atomic ratings points. The antichess abuse is so bad that players have been repeating the same winning GAMES IN THEIR ENTIRETY (to completion!) multiple times against it. At that point, you're not even playing chess.
I think the only solution to this problem is to clear these artificial ratings and prevent bots from playing rated games. I don't think forcing colour alternation is a way of solving this problem. A 1900 player that repeats an exploitable line against a 2500 bot can go 5-5 in a series with relative ease. This would almost certainly not be the case against 2500 rated human opposition.
edit: speling and grammar.
