Currently the ranking in multiplayer bot programming feels a bit unfair sometimes. For example, if a bot has 40% win rate against a particular bot and >80% win rate against all other bots then repeated submits of this bot will cause the 40% bot to become rank 1 on the leaderboard. Even though that bot could have a very poor win rate against most bots. This might seem like an extreme case, but to be more general: after each submit the leaderboard gets shaped according to win rate against you. For this reason you can’t consistently get rank one as long as there is an opponent you don’t do well against, which could be impossible if there is a “rock-paper-scissors situation” among the best bots.
The best way to solve this, taking into consideration both a fair ranking and minimal server costs, would be to run a round-robin tournament between the top 5/10 contestants after a bot achieves the respective ranking on a submit.
Even though the problem is only affecting a small percentage of players. I feel like it discourages people to really go all the way with their bot, which takes away enjoyment from the multiplayer experience.
this is not fair for the rest of contestants (especially for 6-th/11-th contestant)
How exactly is it not fair? With my suggestion the other players will get the exact ranking they would otherwise.
I think this is fair and not discouraging
Do you think it’s fair and not discouraging to get a “random” rank?
Right now the ranking people get after a submit heavily depends on the players they play against and the seeds for those games.
This is not only the case for highly ranked people, but for nearly everyone.
If I’m not mistaken you’ve suggested that top 5/10 contestants should be ranked differently from the rest of the contestants. How can this be fair?
Do you think it’s fair and not discouraging to get a “random” rank?
If I’m not mistaken you’ve suggested that top 5/10 contestants should be ranked differently from the rest of the contestants. How can this be fair?
I’m suggesting an improvement in the ranking system for the people who spent a lot of time improving their bot.
People that are not yet top 5/10 will be more motivated to improve their bot if they will be able to get a much more fair ranking when they do reach top 5/10. So everyone benefits.
As long as there is a large number of random matches the result should be more or less accurate.
It’s unrealistic to run enough games to give everyone their true ranking. This would take a lot of time and server costs.
the problem is that other people’s bots are ranked only once during submission