Points lost to cheats

Originally posted by adramforall
What about those who managed to genuinely win against the "cheat"?

Do they keep the winning games in the league/tourney or lose them?

If they lose the already won games they could effectively lose out on progressing due to other match results.

They lose the points for beating the cheat. They have to, or else the people who lost to the cheat are getting, well, cheated out of points by comparison.

Originally posted by SwissGambit
In tournaments and clan leagues, if any of the players are banned for cheating, their name is crossed out, and their scores zero'ed out.

However, there is no correction to the scores of their opposition. This means those who lose to cheats get zero, while those who still have games in progress with the cheat get wins by timeout.

Instead, all points ...[text shortened]... ] simply by continuing to play until the cheat got banned. The rest had to take losses/draws.

Rec

Originally posted by SwissGambit
I'm not so sure we actually disagree on anything.

The proposal should apply regardless of the reason for banning; it does not advocate changing the rating calculations in any way; it only applies to game-points scored in events [or the rounds of events] that are still active. There is no intent to re-write history, but only to fix things in the presen ...[text shortened]... least he can be stopped from further affecting the results of the tournament once he is banned.

First I think, it does matter, why a player is getting banned.
Getting banned for bad words is something different from getting banned for using engines.
I think, there are three points:
The rating points, the clan chalenges and the tournaments.
All three can be managed as if theese game had never haopened.
The calc shouold be no programming problem

Originally posted by afx
First I think, it does matter, why a player is getting banned.
Getting banned for bad words is something different from getting banned for using engines.
I think, there are three points:
The rating points, the clan chalenges and the tournaments.
All three can be managed as if theese game had never haopened.
The calc shouold be no programming problem

The reason for banning does not change the fact that some people may get free points from a banning, while others do not.

I don't want to see games re-rated, because:
1) It is difficult to program [and that kind of massive number-crunching is a burden on the server].
2) Some cheats are here for years before getting caught, and the re-rate might cause radical fluctuations in ratings [with no obvious explanation as to why, since the bans are 'hidden' now], and...
3) We don't know for sure if the cheat actually cheated in all, or even most, of their games - so a full re-rate might be a big mistake.

I agree with afx that the reason for the banning is significant. A 3(b) ban means that the circumstances of the completed games themselves were suspect. It seems fair, in that case, to strike all results in uncompleted tournament rounds altogether for that player, effectively simply removing him from all the tourneys.

But if a user is 3(a) or 3(c)'d, then the game results themselves were fair. The end result of a non 3(b) ban is no different than if a strong player were to win some tourney games and then resign or timeout the rest due to personal reasons. The wins they recorded were fairly earned and the fact that some other players get "free points" is just the luck of the draw. So I can't see striking all results for a non-3(b) unless we're also going to strike all results for a player who deliberately resigns or times-out in a tourney. Since that sounds kind of silly, I'd restrict your proposal to 3(b) bans only.

Originally posted by SwissGambit
The reason for banning does not change the fact that some people may get free points from a banning, while others do not.

I don't want to see games re-rated, because:
1) It [b]is difficult to program [and that kind of massive number-crunching is a burden on the server].
2) Some cheats are here for years before getting caught, and the re-rate mig lly cheated in all, or even most, of their games - so a full re-rate might be a big mistake.[/b]

at 0: Finnally I understand you. It does not matter, why a player is being banned.
at 1: I am a mathematician and software developer. A little estimation shows, that it would be no problem to recalc. It would cost at most a few minutes to do the recalc. If not a thousands players get banned per day, thats no problem.
at 2: whats the problem about?
at 3: ok, I see your point. There are just 3 possibilities:
1) annulate only games, where the player cheated. Thats not possible, because you don't know, which that where. The cheater could use an engine for just one single move!
2) annulate all, as I suggested
3) annulate none as You suggested.

You pleaded for 3) for a) technical reasons and b) for incertitude
( btw b) is a counter-argument to 2), too )
I countered, that a) is no reason, but did not talk about b).
To b) or not to b), that's a nearly a religious question.
I would prefer 2), but can understand Russ, if he preferes 3).

Whatever you choose, there can not be a fair solution in treating unfair behaviour

Originally posted by Ichibanov
I agree with afx that the reason for the banning is significant. A 3(b) ban means that the circumstances of the completed games themselves were suspect. It seems fair, in that case, to strike all results in uncompleted tournament rounds altogether for that player, effectively simply removing him from all the tourneys.

But if a user is 3(a) or 3(c)'d, th ...[text shortened]... a tourney. Since that sounds kind of silly, I'd restrict your proposal to 3(b) bans only.

But if a user is 3(a) or 3(c)'d, then the game results themselves were fair.

Are you sure? What if the 3(a) user has several accounts entered in the same tourney, and rigs the results accordingly?

And the 3(c) user could be winning by harassing the opposition during games. In OTB chess, you are not allowed to deliberately annoy or harass your opponent. If a 3(c)'er is doing it here, his results should not stand once he's banned.

Originally posted by afx
at 0: Finnally I understand you. It does not matter, why a player is being banned.
at 1: I am a mathematician and software developer. A little estimation shows, that it would be no problem to recalc. It would cost at most a few minutes to do the recalc. If not a thousands players get banned per day, thats no problem.
at 2: whats the problem about?
at 3: ok, I ...[text shortened]... eres 3).

Whatever you choose, there can not be a fair solution in treating unfair behaviour

1) Do you write websites, or just code that executes locally?
2) The main problem here is that people do not want to see their rating suddenly change when no game result has occurred. Players use ratings to enter banded tournaments, clan challenges, equitably against players of their own strength. A sudden, unexpected change in rating skews this. The change could be very big, if provisionally rated games are included.
3) There are known cases of fairly strong human players getting frustrated at playing cheaters and finally deciding to give them a taste of their own medicine. This means, for the majority of their RHP career, they probably did not use engines, and it would be unjust to re-rate all their games as losses. [Their opposition would not deserve the free points, either.] I prefer to err on the side of caution here.

I disagree with your final statement; a 'fair' solution does not require perfect fairness, but only the implementation of practical steps to restore equity, based on the information known.

Originally posted by afx
at 1: I am a mathematician and software developer. A little estimation shows, that it would be no problem to recalc. It would cost at most a few minutes to do the recalc. If not a thousands players get banned per day, thats no problem.

I am an aquarium owner and hat maker.

I don't think you're considering how the rating system works.

If the game by player A (PA) against the banned player (PB) finished 3 weeks ago, then loads of games may have been completed since by PA. A rating "correction" 3 weeks ago would lead to rating "corrections" for every game finished since then by PA. Rating corrections on all games since played by PA would require rating "correction" for all the games finished by PA's opponents against other players since.

It would be an absolutely nightmare, with very little value considering the way the rating works anyway : within 3 or 4 games, your rating is what it should be anyway, despite possibly losing an unfair game.

D

Originally posted by Ragnorak
I am an aquarium owner and hat maker.

I don't think you're considering how the rating system works.

If the game by player A (PA) against the banned player (PB) finished 3 weeks ago, then loads of games may have been completed since by PA. A rating "correction" 3 weeks ago would lead to rating "corrections" for every game finished since then by PA. R ...[text shortened]... , your rating is what it should be anyway, despite possibly losing an unfair game.

D

Yeah, what he said...

Originally posted by SwissGambit
[b]But if a user is 3(a) or 3(c)'d, then the game results themselves were fair.

Are you sure? What if the 3(a) user has several accounts entered in the same tourney, and rigs the results accordingly?

And the 3(c) user could be winning by harassing the opposition during games. In OTB chess, you are not allowed to deliberately annoy or harass your opponent. If a 3(c)'er is doing it here, his results should not stand once he's banned.[/b]

OK, fair enough. I'm not sure how often those two cases have come up in the past, but if a 3(a) or 3(c) can be seen to have affected results of a tourney through multi-use or harassment, it seems OK in that case to nullify all results for that person in the tourney. I just wanted to avoid voiding fair results of games that occurred prior to a 3(a) or 3(c) banning.

In the end, this is really about the effect of 3(b) players on high-profile tourneys and, I would guess, the clan leagues. As long as that case is addressed, how the other cases are handled is not a big deal to me.

Put me in the 'no ratings adjustment' category. It's just too hairy and not worth it for the limited payback it returns.

There is no good solution to this. What is being suggested here is the least bad method of dealing with the problem. So also recc´ed

It would be good to catch cheaters before it makes a difference. Maybe when people get to the last round of big tournaments like this (it just isn´t practical with all of them) maybe a random selection of their games could be run through Analyser II before the final round is started. You could then investigate more thoroughly if it flags anything, thereby hopefully detering engine cheats from joining tournaments in the first place.

Originally posted by Ragnorak
I am an aquarium owner and hat maker.

;-)

Originally posted by Ragnorak
I don't think you're considering how the rating system works.[...]

Just be sure, I have.

I did not think about a very complex correction (you showed, how complex that would be!),
but just recalc all players, all games from the absoulte beginning of rhp.
Lets have 20.000 users (or 50.000 or 100.000, does not matter)
Initialise an array in memory with all users and 1200 points, make a loop over 6.000.000 games (one database cursor), a little calc with
exp and some array accesses, update 20.000 user records in the database.
You can do that with a stored procedure in the database within a few seconds.

( that has nothing to to with local or web apps as swiss gambit mentioned)

Originally posted by SwissGambit
1) Do you write websites, or just code that executes locally?
2) The main problem here is that people do not want to see their rating suddenly change when no game result has occurred. Players use ratings to enter banded tournaments, clan challenges, equitably against players of their own strength. A sudden, unexpected change in rating skews this. The ch ...[text shortened]... only the implementation of practical steps to restore equity, based on the information known.

ok, you convinced me!

the game of cheaters should not have effects to the rating points.
It does not matter, why a player is getting banned.
The games of a cheater within a tournament should be ignored completely.

( ps: I still disagree, that it shall be difficult, to recalc all games.
But it is just not necessary.
)

I definately think that cheats ratings should be absolutely gutted, like in the good old days. That way, at least, the points are back in circulation and can be regained by legitimate players rather then being locked away for the cheat to look and think 'my, what a great rating I got with that banned account, I wonder how high I can get the next one...'