So if practice proves your position, and I flip heads 10 times in a row, that means I will never flip tails? Practice proves that the coin always lands on heads. 
If there's a model that predicts HUGE difference, there'd be some numbers for that in statisticks, don't u think? For now it seems, difference not exist at all (we've not too much data and topology non-optimized), but if we'll follow the logic (there're 100s of 2^N-dices and only 1 winner), difference'd be incredible small at longer distance. I'm comfortable with that.