degaston wrote:Of course there is such a sequence. The most obvious (though unrealistic) example is a data file that contains all the same number:
It is meaningless to ignore the symmetry between the attacker and the defender. That is, yes, every time you attacked from this sequence, you would lose; but every time you defended from this sequence, you would win. The battle outcomes would be significantly different than the expected ones for random chance in this case, but nevertheless on average no one would benefit (which has been the relevant point the whole time). In practice the actual sequence in question was never this far from the extreme, and so while there may have been a shift in the expected battle outcomes I would guess it is small. But the relevant point isn't that, so much as whether there's a bias from hitting some particular patch. And you can only get a meaningful result if you consider that you're likely to be on the receiving end of that bad patch as often as you are to be the one defending from that advantage.