- Updated Friday, April 29th 2016 @ 16:22:42
It would be great to remove some of the noise from the ranking system. Going first or second greatly affects whether a bot will win or lose. What about making all match-ups result in two games with the same opponent where each bot gets to play once as first player and once as second player?
DeveloperCreated Monday, May 2nd 2016 @ 08:51:06
Good point, but this competition isn't really meant to be super competitive, as we also don't have a price for the winner. But overall, in the leaderboard, the results should be fair because on average you will be player 2 as much as player 1.
- Created Saturday, May 7th 2016 @ 06:02:20
True. Although what we see is that in practice the top bot's rating fluctuates by as much as 100 just due to the first player / second player randomness which means jumping from #1 to #5 purely randomly. It would be nice if the ratings were a better indication of the "skill" of a bot.
- Created Thursday, May 12th 2016 @ 14:08:30
A change I would recommend is the positive and negative arrows in the leaderboard. If I understand correctly they display the most recent change? Wouldn't it be better if they displayed the change for each day so that the arrows don't display a random alteration of the leaderboard?