We've been saying for a while that the math didn't look good for Steve Cooley. But now we can go ahead and make it official.
Steve Cooley has lost. Kamala Harris will be the next attorney general of California.
Out of an abundance of
laziness caution, we've held off until now. But we're getting tired of waiting for the AP to call this thing, and Eric Garcetti has already beaten us to the punch anyway.
Update, Wednesday morning: Cooley concedes.
A mathematical explanation/justification after the jump.
The Secretary of State's office has been keeping track of the uncounted ballots on its website. That report lists 394,000 untabulated votes as of this morning, but that's out of date. There are no more than 154,000 ballots remaining, and there are probably less. Update at 2:50 p.m.: With updates from four counties, there are now 124,000 ballots left, at most. (Explanation below.)
The county websites have more up-to-date figures in some cases than the Secretary of State. Using those numbers, we can say that Harris is leading by 49,535 votes. (2:50 update: 51,141 votes.) Given the geographic breakdown of the remaining uncounted ballots, we expect her to expand her lead by another 10,000 votes or so, ending up with a margin of six tenths of a percent (46.0-45.4).
In order to win the race, Cooley would have to win the remaining uncounted votes by a margin of 66-26. (Update at 2:50: Cooley would now have to win by 72-20.) The likelihood of that happening is extremely close to zero. One of his best counties was Orange County, and his margin there was only 60-31.
If we knew nothing about the remaining ballots, you might say that Cooley has at least a chance. But we do know where the ballots are coming from -- they're mostly from counties that favored Harris -- and we know that they're mostly provisionals, which have also favored Harris.
We'd expect Harris to win the remaining votes by a tally of about 50-42. That's just based on geography and doesn't factor in her advantage from provisionals.
We now have much more information about how this race will end up than the AP and the L.A. Times did when they called the governor's race for Jerry Brown at the stroke of 8 p.m. on Election Night.
It's time to call it. Harris has won.
At this point, it's less a question of math than of psychology. How long does Cooley want to go before conceding? Does Harris want to wait for him to concede before declaring victory?
The rest we're going to mark as for nerds only.
A lot of folks -- us, Garcetti, Swing State Project, and no doubt both campaigns -- have expended a lot of effort on projecting ballots county-by-county based on each candidate's performance in that county. But a much simpler and more accurate way to do it would have been to multiply the total number of uncounted ballots statewide on Election Day by 2.5%. That was Harris' margin of victory among ballots cast on Election Day.
When we did that back on Nov. 5, we came up with Harris picking up 58,349 votes on her then-lead of 12,000 votes. All along, that's been a much more accurate guess than the county-by-county estimates, because it ignored geography and focused instead on the character of ballots. As it turned out, the uncounted ballots were more or less evenly distributed throughout the state. The important difference between them and the counted ballots was that they were all cast on or near Election Day.
If you like this story, consider signing up for our email newsletters.
SHOW ME HOW
You have successfully signed up for your selected newsletter(s) - please keep an eye on your mailbox, we're movin' in!
A word on how we arrived at the estimate of
154,000 124,000 remaining ballots. The uncounted ballot report gives the time and date that each county last reported its remaining ballots. But in most cases, counties have added ballots to the count since last reporting their remaining ballots. So we went back and looked at what their tally was when they last reported uncounted ballots, and then subtracted the number of ballots that they've counted since then from their uncounted ballot figure. That's a somewhat crude method, but it's a lot more accurate than relying on the Secretary of State's out-of-date estimate.
Using that method, we can say that the largest chunk of uncounted votes is now in Sonoma County, where they're sitting on 35,500 ballots. Just to be sure that's right, we called the Sonoma Registrar's office and were assured that they're working on it and they do intend to get those 35,500 ballots counted by the deadline. (Harris won Sonoma 57-33.)
Full disclosure for the three or four people who could possibly care: We also estimated that only 94% of remaining ballots will have a vote for attorney general (based on how it looks so far in the counties that give that figure), and that 15% of the remaining ballots will be thrown out because they're provisionals.
Update at 2:50: To the question about why 15%: We had to go with something, and 80-85% is a commonly used estimate for the percentage of provisionals that end up counting. This assumption is a relatively small source of error compared to the error we know is in the remaining ballot tally. If we change it from 85% to 100% -- which would be the assumption most favorable to Cooley at this point -- then the margin he needs drops from 72-20 to 68-24. Still an impossible hurdle. (We can tweak it in the other direction, too. At 80%, it goes to 73-18.)