I have noticed that a decent number of the games I play end up getting cancelled before my opponent makes their first move. This happens especially when I am playing the white pieces. Having to wait 30 seconds for another game isn't a huge problem, but I suspect that that some players are cancelling the game either to avoid playing the black pieces or to avoid the rather unorthodox opening that I usually play as white. For example, I have played 1,040 games as black and only 1,022 games as white.
It's not a major problem, but I bring it up because I imagine that there might be an easy solution. Could we make it so that when a game gets cancelled before the first move, both players end up playing the same set of pieces in their next game? This would not stop players from dishonestly avoiding openings they don't know how to respond to, but it would disallow players from dishonestly gaining odds of the move and should result in players all playing an equal number of games as white and black.
@unchemyst said in #1:
> ould we make it so that when a game gets cancelled before the first move, both players end up playing the same set of pieces in their next game?
this is unnecessary, because it already happens. only it is not completely deterministic. but the more unbalanced your game history becomes with respect to colours, the closer to deterministic it becomes. if you abort black games, at some point it becomes close to impossible to get a white game.
@glbert said in #2:
> this is unnecessary, because it already happens. only it is not completely deterministic.
Thanks, I didn't know this! I think there might be some advantages of a deterministic approach, though. I took a look at the games of my last three opponents: they played, in total, 2490 games as white and 2526 games as black. Not a huge difference, but it seems that my account is not the only one with this pattern. Is it possible that the stochastic method that is currently being implemented to correct for this is still allowing for a small number of accounts with a large excess of games played as white? If these accounts are eventually banned or abandoned because they can only get games as black, the stochastic approach would have no way of correcting for the excess games they played as white. The algorithm that you are describing will reach an equilibrium, but one in which all of the accounts will have played more games as black than white (If all of the accounts that have been around for a while have all played more games as black, the only way to for the algorithm to equalize the discrepancy is to make new accounts play more games as black).
i doubt it. my guess would be that people not using the pool and creating lobby games with fixed colours instead causes more of an imbalance.
The principal mistake is IMHO your assumption that 1040:1022 is a significantly high imbalance which you used as an argument to support your theory that something is not working well. In reality, if you select the color in a perfectly random way 2062 times, the probability that you get closer to equal, i.e. that each color will be chosen 1023-1039 times, is only 29.2%.
@mkubecek said in #5:
> The principal mistake is IMHO your assumption that 1040:1022 is a significantly high imbalance which you used as an argument to support your theory that something is not working well. In reality, if you select the color in a perfectly random way 2062 times, the probability that you get closer to equal, i.e. that each color will be chosen 1023-1039 times, is only 29.2%.
Against that, of course, is the fact that colour selection is *NOT* random but determined by previous history. However, your point is valid in that the observed bias towards Black is indeed very slight.
Right, with a system that tweaks the probabilites in order to balance better, one can expect better balance than for completely random selection. However, the fact that this sample is in fact balanced better than average could be seen as an indicator that there is no need to apply any corrections yet.
> my guess would be that people not using the pool and creating lobby games with fixed colours instead causes more of an imbalance.
As a side note, I always found it a bit strange that you can create a *rated* game with the color of your choice.
Just out of curiosity: if I created 10 games in a row where I pre-selected White, would I be more likely to get Black in the next several random-color games?
@Frogster64 said in #8:
> [...]
> Just out of curiosity: if I created 10 games in a row where I pre-selected White, would I be more likely to get Black in the next several random-color games?
Yes, I believe you would, assuming you started out with equal numbers of white and black games before those 10 white games. As I understand it the algorithm attempts to make the total number of your white and black games the same. [We're talking about non-tournament games of course. Certainly Swiss tournaments do their own colour selection independently of a player's previous history. For Arena tournaments I don't know how colour selection is done.]
Some people play more often with Black because they participate in simuls.
