The polls posted on RCP today show Obama ahead by 5, 5, 8, 5, 3 and 3 (plus the odd Gallup "expanded" poll, showing 10). So let's say the median is 5, which is closer than other recent polls. And except for the Gallup expanded, the most Obama gets in any of the polls is 51%.

So here's the scenario: if McCain is in fact behind by about 5 points (a big if), he has eight days to get some enough Obama-leaning voters and undecideds to vote for him to narrow the gap to two points; difficult, but a three-point swing, including the 4-9% of voters who are undecided, is not out of the question. With regard to undecideds, some have argued that the real "Bradley effect" is that undecideds when there is a black candidate tend to break for the white opponent. (538.com disagrees).

Once the gap is that low, a narrow electoral majority is possible. 50,000 or so more votes for Kerry in Ohio in 2004, for example, and Kerry would be president, even though he was behind by 2.5% in the popular vote. With big electoral vote states like New York, Michigan, Illinois, and California going overwhelmingly for Obama, and many traditional Republican states pretty close, the popular vote estimates could overstate Obama's electoral margin considerably (or understate it considerably, if he wins all the close states).

Likely? No? Impossible? No. Of course if the polls showing a 3% gap are correct, then the race may already be something close to a tossup at this point. And if the slightly older polls showing a gap of nine or more points are correct, McCain's chances are closer to none than to slim. Likewise if Obama's vaunted turnout machine is as good as some claim, and the polls are undercounting Obama voters.

UPDATE: Just to be clear, I think the odds heavily favor Obama. But there's a difference between being the heavy favorite, and a sure thing.