"Two players with nearly equal skill would need years of playing to know who the best is for sure. "

Not sure I agree with this.

It's like roulette - the house odds may only be 5% but they're 5 % every single spin of the wheel. Your 95% return per spin means you actually win 0.95 to the power x, where x is the number of spins you play for. After 20 spins, you're down to 35% of your original stack.

I am suggesting that you can take the same conecpt to poker (though *measuring* the difference in skill will only be possible after the fact, and because human error is involved in a way the roulette wheel will never suffer from, getting workable empirical data might be near impossible). Even a tiny recognised advantage one player has in skill will end up snowballing in the same way. Luck will play a huge part in terms of variance; but in a relatively short time, the better player will end up considerably up over the worse player.

Thoughts?