The Democratic rules provide for delegates to be awarded proportionately on the basis of the popular vote. It wasn't always that way, but a change designed to weaken the control of party bosses was ushered in after the riotous Vietnam War-era 1968 convention [...]

In a race with two equally matched rivals -- Obama and Clinton are both running well-funded national campaigns -- that tends to leave the winner of the popular vote with only a narrow delegate advantage over a loser who runs a strong race.

Multiply that across dozens of congressional districts -- 53 in California -- and predicting the winner of the delegate struggle is a virtual impossibility.

Then it gets harder.

For the Democrats, in a congressional district with three delegates, two go to the popular vote winner, and the loser gets the third as long as they win 15 percent of the popular vote.

But in a congressional district with four delegates, the winner and loser in a two-way race are likely to divide the spoils evenly. The winner must receive nearly 63 percent of the vote to get a 3-1 split in delegates, and 85 percent of the vote to win all four.

This is generally very true.  But the author neglects to mention that there are only two Congressional districts in California which offer 3 delegates.  The real prizes are the five-delegate districts, because the majority of the districts offer even-numbered delegates which almost guarantee an equal distribution.  And because of the particular breakdown of delegate allocation, Barack Obama actually has a built-in advantage in winning a majority of the delegates, regardless of the popular vote.

Here's the list of delegate allocation in California.  As you can see, there are 370 pledged delegates up for grabs, and 241 of them will be pledged at the district level.  The other 129 will be allocated to candidates based on their share of the statewide vote.  Given what we know about the closeness of the race in California, I simply can't see much more than a 10-point spread in that allocation.  So the other 241 from the district races will end up being a significant factor.

There are 26 districts which allocate four delegates, all of which are almost certain to split evenly among Clinton and Obama. Significantly, these include some of the most heavily Latino districts in the state, including CA-21 (Nunes), CA-31 (Becerra), CA-32 (Solis), CA-34 (Roybal-Allard), CA-38 (Napolitano), CA-39 (Linda Sanchez) and CA-43 (Baca).  Clinton's perceived advantage among Latinos is neutralized by the high bar needed to cross to gather extra delegates in these districts.  The likely scenario is an even 52-52 split.

There are 6 districts which allocate 6 delegates, where it is still likely to be an even scenario, but where a strong showing could give a 4-2 split (I think a candidate would need close to 60% of the vote for that to happen).  These districts, the most Democratic in the state (the allocation is based on Democratic turnout in primaries), are CA-06 (Woolsey), CA-08 (Pelosi), CA-09 (Lee), CA-12 (Lantos), CA-14 (Eshoo) and CA-30 (Waxman).  As these are districts populated with liberals, and given that some of them are high-income (06, 08, 12, 30), they seem to trend toward Obama.  I think CA-09, Barbara Lee's district serving heavily African-American Oakland as well as some other East Bay cities, offers the best chance for a 4-2 split.  Let's say that Obama gets one of these.  The number is now 71-69 Obama.

As I said, there are two districts with 3 delegates: CA-20 (Costa) in the Central Valley, and CA-47 (Loretta Sanchez) in Orange County. (As an aside, this means that these two districts turn Democrats out to primaries at the lowest rates.  And they both have Democratic Congressmen.  Way to go, Bush Dogs!)  I project that CA-47 will go to Clinton, and think that CA-20 is up for grabs.  There are a decent amount of campesinos in that area, but rural districts in Nevada went strongly for Obama.  So let's hold off on that for now.  The number is now 72-71 Obama, with 3 delegates outstanding.

Now we come to the real electoral prize: the 19 districts which offer 5 delegates.  There are quite a few advantages for Obama in these districts.  First, all three heavily African-American districts in Southern California are in this group: CA-33 (Watson), CA-35 (Waters) and CA-37 (Richardson).  Obama should be able to attract a majority here.  Then there are two districts in the far north of the state: CA-01 (Thompson) and CA-04 (Doolittle).  Based on how their Nevada neighbors voted, I project them to Obama.  Third, there are three districts in the Bay Area that fall into this category, and in the most recent Field Poll, Obama was stronger in the Bay Area than Southern California.  I expect him to take CA-07 (George Miller) and CA-13 (Stark), but lose CA-10 (Tauscher) because that's a more suburban district.  That's so far a 7-1 split for Obama.

Clinton's strength is in the suburbs and in Southern California, as well as among Latinos.  But very few of those districts fall into this grouping.  There are three in the San Fernando Valley: CA-27 (Sherman), CA-28 (Berman) and CA-29 (Schiff).  But Adam Schiff has strongly endorsed Obama, and his Pasadena district is more liberal and upscale.  I see a 2-1 split for Clinton here.  NONE of the Orange County districts offer 5 delegates.

Going into the wild cards, we have 8 districts for Obama and 3 for Clinton.  The rest include CA-05 (Matsui) in the Sacramento area, CA-15 (Honda) in the San Jose area, CA-17 (Farr) in Monterey, CA-23 (Capps) in Santa Barbara, CA-36 (Harman) in the South Bay of Los Angeles, CA-50 (Bilbray) in the San Diego suburbs, and CA-53 (Davis) in San Diego.  If I were to guess, I'd say that CA-23 and CA-36 have some built-in advantages for Obama (upscale, highly educated, "wine track" liberal), making it an 11 to 3 split, with 5 outstanding.

So, before the polls close, we can reasonably project a 111-102 split for Obama, with 28 delegates up for grabs, as well as the 129 that will go proportionally to the winner.  If you split the rest of the district-level delegates evenly, I think you end up with anywhere from a 7 to 12 delegate advantage that Clinton would have to make up in the popular vote.  At the lowest level she would need 53% of the vote or a 6 percentage-point victory to make this up; at the highest level, 55-56% of the vote or a 10 to 12-point victory.  Given the polling recently, and the fact that there has been an unusually slow rate of return of absentee ballots until after the South Carolina primary, I think the final result is likely to be narrower.  And so, despite the possibility of Hillary Clinton winning the popular vote in the Golden State, I'm not sure she's favored to win the majority of delegates, given the reality of the allocation.

Now, the question becomes, how will this be spun?  Will the media only report on the popular vote, or will they look at the delegate counts?  Probably the former; it's simply easier for the format of broadcast news.  But they'd be missing out on an important story, that this is a race for delegates, and the candidate who takes advantage of the system is the most likely to reap the benefits.