[AndyTheBear]

Which makes me wonder. Isn’t it usually the case that early returns favor Republicans and later returns favor Democrats, because big city precincts take longer to count? That could be a flaw in this method.

Yeah, that is usually true. With only 4% in we aren't going to know if its above or below the tipping point unless we have tipping point data for the particular areas in the 4%.

[Stat Man]

Seven states now with partial results in the spreadsheet I made based on this blog’s idea.

Shows Romney 29% chance of winning vs 71% for Obama.

Romney trailing early in FL and NC main reasons why.

Like you say, I think this is probably flawed, depending on where the early votes are coming from. Once a lot of states are beyond 50% of precincts in, this system probably starts approximating reality and becomes a lot more useful.