It's been obvious for a while now that Hillary Clinton's ability to continue her candidacy to the Democratic Convention depends on her success in at least one of three dimensions (assuming she can't catch Barack Obama in pledged delegates without some complete collapse of his campaign): (1) an uniterrupted winning streak in the final primaries; (2) the sudden appearance of a major positive differential in her general-election poll standings from Obama's; or (3) a plurality in the final cumulative popular vote. Number (3) is the most plausible for her right now, particularly if she can convince the media and other Democrats to choose a measurement of the total popular vote that gives her the best chance of catching up.

This last note reflects the little-comprehended fact that there's not any sort of "official" tabulation of the popular vote; nor will there be one when the whole game's over.

Pollster.com's Mark Blumenthal has an analysis up today on the National Journal site that runs through some of the difficulties involved in figuring out the people's choice. They go beyond the question of whether votes cast in unsanctioned primaries in MI and FL ought to "count," along with the fact that popular vote totals aren't available for four caucus states. There's also the anomalous situation in WA, which held a "non-binding" primary to accompany its delegate-selection caucuses.

Blumenthal also raises the technical but politically explosive issue of "measurement error:" at some point, the margin of victory or defeat for a candidate in an extremely close race becomes smaller than the number of votes cast or counted in error.

The bottom line here is that there's enough confusion about popular vote totals to enable HRC to at least attempt to claim a victory if she can come up with a version of the results that passes the laugh test. And that could lead to a "numbers war" that will be very difficult to resolve.

For that to happen, of course, she needs to keep winning primaries, and the underlying big question is whether superdelegates will tilt decisively to Obama if he can win in, say, North Carolina. I think that's more likely than not.