Again, we both did multiple trials, and none got the same outcome >70% of the time. Your most extreme trial came close, yet we cannot say, "the most extreme of double digit trials ALMOST matched the poll results; therefore the polls suggest no advantage." Within a minimum of 10 trials, the upper bound of inequality was 69%, less than the polls (which have 36% more samples than each of our trials.) Therefore the collective polls most likely show(ed) a genuine Obama advantage.

Granted, that last statement merits MANY qualifiers; none of the polls shared the same sample, and many had different methodology/weighting, so they are analogous to flipping and catching quarter, then flippin a nickel and letting it bounce, then spinning a dime on its edge, etc. until we reach 136. However, as long as each different coin/toss is fair, other differences are irrelevant; the same probability is constant. And, of course, the Law of Large Numbers again tells us that as the number of polls increases all inadvertent biases toward either candidate correspondingly diminish.

Quantifying the size and strength of Obamas advantage in pre-debate polls (and preliminary polling since suggests that advantage took a BIG hit Wednesday) remains impossible, particularly since the victory and error margins were usually very close. However, it is highly improbable (though possible) the collective polling represents NO real advantage for Obama. How improbable is, of course, also impossible to say.