So I decided to torture myself by looking at a few different ways of tallying the popular vote. Since there's no concept of the national popular vote under Democratic Party rules, there's an unlimited number ways you can slice the pie. Thus, one of the many problems with using the popular vote as meaningful measure is that nobody will ever agree on how to count it. (Here's some other problems.)

Anyway, here's five different ways you could count the popular vote, some more reasonable than others. (I didn't waste my time considering the methods that exclude caucus-only states.)

I'll bet you can guess which ones I favor...

Also, don't forget that over the past month, McCainiacs have cast 362,000 votes for Hillary Clinton. Without their votes not even the most creative math would show her leading the so-called popular vote.

Methodological note: TX stands for the Texas caucuses and WA is the Washington state primaries. I estimated the vote distribution the TX caucuses at 605,000 for Obama and 495,000 for Clinton. To determine Obama's Michigan vote, I used the most conservative possible estimate, using exit poll data that indicated 79% of of the 35% of voters who would have voted for Obama had he been on the ballot cast their ballot for uncommitted, yielding a total of 164,351 votes. According to the exit poll, Clinton gained about 40,000 votes from Obama supporters, but I did not subtract those from her total.