>>13156750That is a very bad example. Chess is not a 50/50. You would have to be randomly sampling over "even" matches, which itself is completely undefinable. Are chess matches between the two best players in the world even fair? Given that chess is unsolved and appears to be a skilled game, there is never a 50/50 split as the better player is more likely to win. So they are tracking the likelihood of the better player player to be white or black. This seems like a 50/50, but would this actually be reflected in wins / losses? What about draws?
Rules, like those in tournaments, offer alternative win conditions which makes things like draws preferable to one player or outright losing to another. If half a point means a players point lead is guaranteed first going into the next match, then a draw is a great outcome. The player is then more likely to choose a drawn end game if presented. So same game series will have higher likelihoods to draw. Even if draws are ignored, that means the preference is carried over to missing games for some color.
So the aliens must be sampling blind to players(who they are) and events(when they are). If they don't then their model will have inconsistencies that want an explanation.
Is ignoring draws even a fair option? The current idea is based on a first move advantage and so better players as black would be happy with draws. So even in that data there would be a ratio discrepency if draws are ignored.
