crimes of quantification

Slate is running an article on how the current poll numbers look more favorable for Kerry than what others are suggesting. The article makes several good points. But it also contains this strange lapse in reasoning:

"Look at the numbers for Kerry in these [polls]: 50, 48, 49, 49. Even in the CNN poll, he's got 47. Kerry is that close to making a Bush victory mathematically impossible."

A Bush victory is mathematically impossible if only Kerry can secure 50.1% of the votes? If only America decided its president by that quaint means known as "majority rule." As it is, national poll numbers really don't go very far in saying who is ahead in the presidential race.