Jonathan Chait must be busying himself with some historical data-mining; he found the pre-election prediction of RealClearPolitics in 2000, and it's hilarious. The site's best guess was a 10-point Bush landslide, with Ralph Nader at 5.7 percent, and a 446-92 domination of the electoral college. Even Dukakis got more than 100 electoral votes.
For those who still maintain [sic] Mr. Gore has a chance of winning, consider the scenarios under which this is possible. If Gore does not win Florida (the evidence indicates he will not), he must run the table, taking IL, CA, PA, MI, MN, WI, WA, OR, TN, AR, WV and DE along with his base 92 votes for a 273-265 EC win. It won't happen.
Not only did Gore win all but three of those states, he won most of them easily. He took California and Illinois by 12 points, and took Washington, Michigan and Pennsylvania by 5. Wisconsin and Oregon only stayed close because of a solid Ralph Nader vote, but Nader won less than half what RCP predicted. The swollen Nader polling was one reason Gallup failed to predict that Gore would win the popular vote, and most pollsters made the same mistake. A heartbreakingly close presidential election helps turn people into experts on vote counts, eventually.