This item in the Wall Street Journal suggested to me that the Bloomberg third-party-bid talk is a giant head fake:

The Empire State might prove less hospitable than other large electoral prizes because more voters there have well-established partisan ties. A better target: California, where proportion of voters who “decline to state” party affiliation has more than doubled to 19% since 1980. Dominance of TV ads in California would also maximize leverage of Bloomberg’s checkbook. But disparate profile of swing voters complicates the New York mayor’s possible challenge. Though his prospects are strong among upscale independents with moderate social views, he’d face uphill climb among populist, socially conservative blue-collar independents. “For 80% of the voters,” notes Democratic strategist Bill Carrick, “the most important information” remains candidates’ party affiliation.

In other words, Bloomberg would clean up among independents and socially liberal voters, and leave the conservatives. He would hurt the Democrats in California, and potentially cost them 55 electoral votes.

George W. Bush got 44.4 percent of the vote in California last time. In other words, if the Bloomberg’s push peels off about 10-11 percent of the electorate out of the Democratic pool (and minimally out of the Republican one), the GOP could sneak away with the first win in California since 1988. I just don’t think Bloomberg, who can pretty safely be described as a liberal, would want to do that.