UPDATE: IT’S OVER! PEREZ CONCEDES! (Was: Can Perez Still Win Ugly? Here’s How He Might Try)


Parke Skelton, Strategist for the Betty Yee for Controller campaign, has notified OJB that Perez has announced that he is abandoning his recount and endorsing Betty Yee.  This is not a drill — it’s over!  Congratulations to Betty Yee for her victory and to John Perez for doing the right thing sooner than might have been!
Just a reminder: if we simply have a simultaneous hand count of all ballots in Glenn, Humboldt, Imperial, Kern, Mendocino, San Mateo, Santa Cruz, Shasta, Siskiyou, and Yuba Counties — a total of about 306,000 votes out of 4,040,000 cast state wide that match the profile of the ballots cast statewide — then within a couple of weeks we’d probably all be able to agree on whether Perez (or Yee!) still has a chance to win.)  This could be arranged by agreement between the campaigns and is both very fair and far cheaper. than what is happening know.  The proposal presumes that disqualified ballots in these counties are in proportion to those in the other 48 counties — but, even if that presumption is wrong (and under Perez’s demanded recount it would never be tested anyway!), the speed and informational value of such a preliminary count would be very useful.  It would leave Perez secure that, if he did withdraw, he truly had no chance to win.
Cherries in cupped hands

By choosing only some cherries — by which, for the overly literal, I mean PRECINCTS — for a recount by HAND, yet subjecting the rest to an inferior MACHINE count, Perez hopes to catch Yee napping.  If she waits until Perez finishes counting Los Angeles, it might be too late for a counterattack.

NOTE: UPDATES will appear throughout the text, where appropriate, in orange.  I’ll note their existence up here, as so:  “UPDATE 1 just posted 7/17 4:15 pm.”  (No kidding, it’s really down there: Perez abandons Kern count!)  Search on the “UPDATE [#]” to find the latest updates.

1. The Background

The recount of the votes — that is, some of the votes — in the Controller’s race between John Perez and Betty Yee (who led Perez by 481 votes out of almost 4,040,000 at the outset, the closest statewide margin between two candidates in California history) begin in earnest this week after a “slow start” on Friday.  Some of the stories to tell about the recount are mathematical, some are purely political, and some are legal — and most depart substantially from the “Perez can’t win” stories being written today.

I think that Perez can still possibly win.  But his best chance to do so would require aggressive and underhanded legal tricks.  What some commenters don’t seem to understand is that, theoretically, those tricks could work.  But that short-term victory for Perez might be a disaster for the Democratic Party otherwise.  (Republicans, of course, should be rooting for him to do exactly this.)

Perez has asked for recounts in 15 counties, all of those where he won his one-on-one competition with Yee.  (Ashley Swearingen actually won 10 of those counties, to Perez’s 5.)  He asked to count the first two counties on his list — Kern and Imperial — simultaneously, and then the rest sequentially.

2. The Calls for Perez to Stop the Count

The past few days have seen a crescendo of attacks on Perez for not stopping the count.  For example, from Calbuzz, there’s Why Bully Boy Perez Should Concede to Betty Yee:

Former Assembly Speaker John Perez is a bully. He was a bully as Speaker and he’s a bully now, weilding his financial advantage over State Board of Equalization member Betty Yee to use California’s screwed-up election law to cherry pick precincts in hopes of overcoming the 481 votes by which he took third place to his fellow Democrat in June’s election for Controller.

When he announced he would be seeking a recount, he issued a statement about how important it is “to ensure that every vote is counted” and how he had made “the defense of voting rights a core part” of his career.

What a bunch of self-serving clap-trap. Perez isn’t interested in ensuring that every vote counts – he wants to manipulate the vote count in the precincts where he performed best in hope of overturning the tally and making the November runoff against Republican Ashley Swearengin, the mayor of Fresno.

Yee, as decent and intelligent person you can find in politics, has been remarkably restrained in her reaction, and in the meantime has picked up the endorsement of the California Democratic Party — along with some much-needed cash — and won the respect of myriad others who see her as the actual second-place finisher.

His supporters argue that he’s only doing what the law allows and that it’s every candidate’s duty to fight to win. Fine, but if you use a broken – and likely unconstitutional – law to selectively recount the vote where you think it’ll do you the most good, just don’t insult the voters by claiming it’s for some civic-minded cause.

[Quoting Dem consultant Garry South:] “There’s no doubt the California law on recounts is completely screwed up and needs to be totally rewritten. But it’s another thing for a defeated candidate to try to take political advantage of the weaknesses of that law by covering it with high-sounding rhetoric about an obligation to make sure every vote is counted — but only in the counties and precincts where he won.”

I am an ardent Yee supporter, but I agree with Scott Lay who points out that an effective Speaker is generally a bully.  More to the point, Yee could, spending a minimal amount of money, put an end to the malarky right now if she wanted to, either through selective recounts of some small counties that favor her or through a more aggressive and spectacular tactic that I won’t yet describe publicly.  Yes, South is right about the ethos of underhanded tactics, but the scorpion’s gonna sting and Yee has some responsibility to take appropriate actions in response to that threat.

My “Comrade in Controller Coverage” John Hrabe, writing in CalWatchDog (but often appearing in CalNewsroom, Fox&Hounds, and elsewhere), has offered a similar but more mathematical take in Controller 2014: 7 reasons why John Perez should quit while he’s behind:

1. His current 15-county recount won’t be enough.

More important than the current tally is the current error rate, which isn’t high enough to overturn the results. Two days into the recount, 91 precincts have been recounted with Perez gaining at most five votes (Kern: +4; Imperial: +1). Although the sample size is small, the current rate has Perez gaining one vote for every 18 precincts (18.2).

Perez’s recount request only identified 4,103 precincts to be recounted. If our math’s right, after those precincts are recounted, Perez could expect to gain just 225 votes — not enough to change the results. (4,103 precincts / 18.2 current change rate = 225.43 vote gain for Perez)

Hrabe’s case has, if anything, gotten better in the days since he publishes.  Buts I’ll note in the piece, he’s missing some important items in this analysis.  Here’s a previous of what I have to say on the topic below.

(1) “Precincts” don’t matter.  Comparative margin in those precincts matters.

The statistic we should be looking at is not the number of precincts outstanding, but a conversion rate that, so far as I know, doesn’t have a formal name.  Perez has chosen to count precincts where he leads Yee — in Kern, by even one vote; in Orange County, by at least 8.  The principle behind this is pretty sound in helping to guide what is in any case a stab in the dark:

Whatever factors led people in this precinct who have had their votes counted to vote for Perez are also likely to have led anyone in this precinct whose votes haven’t been counted to vote for Perez.

But that just suggests where to find fertile ground for errors in one favor.  It doesn’t tell you how likely you are to find such errors.  That’s where we need our “conversion rate,” which is essentially “how many ballots does one need to examine to find one favorable error,” which I’ll abbreviate as “Ballots Required to Add One Favorable Vote,” or BROAFV.  (I put it in red so that you can find this section once I start using this term below.)

One nice thing about BROAFV is that it only has to be applied to the margin between two candidates, not to their vote totals.  If BROAFV indicates that you should get one favorable vote for every 100 ballots recounted, than a precinct that one candidate wins 200 to 100 has the exact same expected effect as one where the candidate wins 500 to 400, because statistically the expected gain from 100 ballots or 400 ballots common to each candidates total would cancel out.

Precincts differ in size.  Votes don’t.  So to figure out how likely a candidate is to switch a certain number of votes, you don’t need the number of precincts at all: you just need (1) the total aggregate margin by which the candidate leads in those precincts and (2) the BROAFV.  Multiply the two and you get the number of votes that the candidate should be able to expect from that set of precincts.  (Of course, the actual number obtained will vary due to random errors — as well as possibly by systematic biases.

(2) Perez can add precincts from any of the 15 counties to the hand count at any time he pleases

It’s not quite accurate to say that Perez is only recounting ballots in his priority precincts.  By law, he has to recount every ballot in the county for that count to affect the recount.  But his argument is, and my information from the OC Registrar of Voters suggests that they agree with him, that he can apply different counting methods — a hand count that is likely to turn up any wrongly rejected ballots, and a machine count that is almost certain not to do so — to different precincts within a county.  (I think that the law on that point is ambiguous — and that Perez’s interpretation is an unconstitutional violation of Equal Protection.  What he’s doing right here is itself the best argument for why that is!)

So, while Perez has only identified 4,000 or so precincts for manual recounting, he could double or triple that number any time he wanted to.  He has explicitly reserved the right to do so in his letter to the Secretary of State — and apparently adding more precincts to a hand recount is legal.  (I’m less sure that subtracting them from the hand recount — that is, “cancel that order to hand recount Lake County and do them all by machine” — is legal, but I’m sure that Perez thinks that it is, and that they have a viable case to make on that point.

Any precincts added, of course, would be inferior to those already counted — which is Hrabe’s next point.  But as I’ll show below, those less-good counties still hold some significant value.

Hrabe continues:

2. Cherry-picking means the error rate is likely to get worse for Perez, not better.

At the start of the recount, the focus was on Perez “cherry-picking” counties and precincts where he outperformed Yee. “How unfair,” everyone bellowed. Now, the cherry-picking works against Perez. If the current error rate holds, Perez must expand the recount (See Point #1). But the error rate has nowhere to go but down. If he expands beyond the 15-county recount universe, he’s likely to lose votes, or at least see a lower rate of change.

He’d never be expected to lose votes, so long as he sticks to precincts where he leads and the assumption underlying counting favorable precincts holds true.  (He might lose votes in practice because that’s only an expect result, and doesn’t count random error.)  He would likely see a lower rate of change, though.  Still, if those uncounted precincts are smaller — something I haven’t checked — then he might be able to expect the same bang for the buck.  (Counting five 30-20 precincts should take about the same about of time as counting one 150-100 one, but the latter would have been much higher on his list.)  Again, we don’t look at the number of precincts, but at the aggregate vote margin within them.

3. Why Perez Probably Won’t — and Even Shouldn’t — Stop the Count Yet

If and when Perez stops the count, it will probably be on a Friday afternoon because that’s when one wants to put out the bad news.  So it could come tomorrow.  But I don’t think that it will — and if I were his advisor, wedded to a plan that Garry South would consider unethical, I likely wouldn’t recommend it.

Regardless, I don’t think that it makes sense for the Yee camp to call on Perez to stop the count until it’s clear to his supporters, whose votes she will need this fall, that the game is truly over.

Even if Perez is a bully and so on, his supporters do believe in him — and the belief that Yee is playing fair with him (even if he isn’t reciprocating) is important to her success.  So my interest in this piece is to assess, with what I think is a bit more sophistication than I’ve seen elsewhere, when and how Perez could and should decide that that tantalizingly small victory margin Yee has really would hold up.  So I’m going to summarize what I think his best, nastiest, strategy would be.

If you want the bottom line — I think Perez’s effort will fall short, probably, unless the unexpected awaits us in Los Angeles, by more than 300 votes.  But I think that, at least until Monday, August 11, Perez should allowed to go forward with minimal criticism.  So long as Yee is kept in the news as a gracious and impressive figure, I don’t think that her campaign is injured by another 3½ weeks of recounting, during which her touring the state is more likely to make the news.

Now I’m going to put on my “fake Perez advisor” hat and try to figure out what I’d tell him to do.

4. Amoral Advisor Here, at Your Service!  (Relax, It’s Just Play-Acting) 

I’ve tried to take the role here of an amoral (but legally ethical!) election law attorney, advising Perez on what he should do.  Specifically, could he win by continuing a hellaciously biased recount — a recount whose intricate bias is, in the manner of a lion hunting down and devouring a zebra, in its own terrible way somehow a thing of intricate beauty?  After all, for unfathomable reasons what he has demanded remains, for now, within the boundaries of legality — and the Amoral Attorney Code says that that’s all that should matter to me in giving him advice.

I started writing with a clear bias: yes, if he pursued an aggressive cherry-picking strategy, he still could win!  You’ll see where I ended up.

5. The Clock

The weirdest thing, from my perspective, about the structure of the Perez recount is its sequential nature once the first two counties are completed.  Let’s go back to John Hrabe’s piece one last time to discuss his third reason why Perez should quit now:

3. There’s not enough time to sequentially recount.

In addition to the error rate, Perez could fall short on time. According to a survey of county election officials, the recount — at its current pace — would stretch past the November general election. As first reported by the Sacramento Bee’s Jim Miller, “A new survey by the secretary of state’s office suggests that the recount could run through January if it includes all of the counties sought by Pérez.”

If Perez begins to pick up votes, Yee’s campaign could at “any time during a recount and for 24 hours after it concludes” request her own recount, “as long as it does not include any precincts that were recounted as part of a prior request,” according to the secretary of state’s office. 

Perez’s approach seems like madness, given that linear extrapolation that the recount will extend through January.  Not only does voting start in early October and extend to early November, but the first ballots have to be printed by September 5, meaning a recount finishing, most likely, by around Friday, August 29 — or, if someone works weekends and Labor Day, Sept. 1 or 2.

Ha-ha, stupid Perez, huh?  Nope, not at all.

Obviously, running out the clock is part of Perez’s strategy or he wouldn’t have done this!  His Hail Mary pass here depends on his going into the lead and finishing his recount so late that Yee cannot start her “counter-recount” in time.  Of course, she could start it now — I’ve been publicly urging her to do so, just to show Perez that she’s aware of his ploy — but she has shown no signs of doing so, as she wants to save her money.  That $50,000 she just got from the California Democratic Party (CDP) won’t actually go very far if she wants to start counting her best counties such as San Francisco, Marin, Mendocino, Monterey, Santa Cruz, and Sacramento.  To pick up votes, she’ll need a manual count — and hand counts take a lot more time.

If Yee wants a recount to enter the final tabulations, it has to be entirely completed — as Perez’s will be, one way or the other.  At any time, given certain legal assumptions discussed above, Perez can take a shortcut towards completing his recount within a day or two, if at some point — probably after Los Angeles is completed, he manages to pull ahead.

And this answers the question of why Los Angeles is ranked only #8 on the list, behind counties where Perez did worse, such as Orange at #6.  (Los Angeles was actually Perez’s best county in terms of vote margin — and for a selective manual recount vote margin is the main thing that matters.)

Perez would probably like Los Angeles to finish by around August 20.  That timing creates a problem for Yee: to get her recount safely done without having to switch to useless machine counts, she should probably start by August 15.  If Yee has fallen behind by perhaps 70 votes on August 20 or so , it may be too late for her to call for even a simultaneous full manual recount in most counties; Kern, for example is taking two weeks.  Yee will have to gamble — expensively — on doing many of them simultaneously or else she will have to restrict her recount to fewer counties.  The danger there is: what if she chooses counties that did everything right?  That’s why she should get at least some of the easier to count ones — Mendocino, Sonoma, Marin, Monterey — out of the way NOW.  Not only may she put the election out of reach and lead Perez to quit early, but she can then call for more selective recounts in San Francisco, Sacramento, and maybe San Diego and others.

(Why wouldn’t she count the favorable precincts that Perez excluded from his “priority precinct list” in Los Angeles, Orange, and such?  Read the Election Code.  I may discuss this further next week.  It’s the main reason that Perez will lose an equal protection challenge — which I’ll bring myself, if no one else will.)

Perez needs to get lucky not only with the recounts in San Bernardino, Riverside, and especially Los Angeles — but then he needs a court that will be willing to cut short the process even if everyone will agree that Betty Yee probably got more votes in the primary.   If her recount starts too late to do its magic, a court could side with Perez under the legal principle of “laches” — which is, in essence, the principle of “you snooze, you lose.”  Once Perez truncates his count at pretty much the last possible moment, he’ll argue that the Sept. 5 deadline for sending ballots overseas is inviolate — or at least that it should not be violated if Yee “slept on her rights.”  (There’s another reason that I argue that Yee should start her recount of at least small counties like Mendocino and Sonoma right now — to show that she’s not sleeping on her rights.)

The prospect that Perez might win such a case takes him out of “Perez can’t win” territory — in which he’s stuck if he does stop counting — so long as he has a plausible pathway to producing the votes.  That pathway becomes a lot easier if Yee can’t push through her own selective hand recount.  (Of course, I have to prove to you that Perez can still pull ahead at some point.  Be patient!)

But one last thing: how will Perez time the end of his recount so accurately as to ?  That part is easy.  He’s only asking that certain priority counties in the 15 identified counties be manually recounted — those, naturally, being the ones where he leads.   The rest, in each county, can be counted by whatever machine method was originally used to tabulate them.  That massively reduces the total time of the count — and probably freezes its numbers where they are, because machine recounts rarely move numbers much at all.

Let me use the example of Orange County — #6 on the list of 15 to be counted — to explain how that works.  In OC, Perez has identified 516 precincts that he’d like to recount manually.  But OC has 1856 precincts overall!  About 104 of them are either empty “ghost precincts” or nearly so, leaving us with 1752.  To make the math slightly easier, let’s say that 32 other precincts aren’t ghosts, but nearly so, leaving us with 1720 that might make a difference.  That means that Perez is asking to count exactly 30% of OC’s precincts by hand.

I’m informed by people at the OC Registrar of Voters that they’d estimate 7 days for a complete manual count and about a day for a machine count.  So, that’s a little over 2 days for a manual count well under a day for a machine count of the rest — and 7 days of counting just turned into 3 days!  This truck can be used with other counties as well.  (My guess is that, if time is short, counties #9-15 might be shunted into machine counts en masse before Los Angeles is finished, unless they look like they may be decisive.)

My estimate is that if one calculates the timing of the precincts to be counted with the proviso that Perez may cut some counties short by moving prematurely to a machine count, he’ll finish pretty much as close to the final moment — sometime between August 28 and September 2 (between which is a three day weekend) — as he wants.

6. How Do We Best Extrapolate from Current Projections?

In Imperial County — Perez’s “best county,” as Yee’s strategist Parke Shelton put it, where he led Yee by 41.4% to 10.2% of the vote (with the other 48.4% divided up among the four other candidates) — Perez has picked up only 5 votes of the 481 he needs to tie.  Woo-hoo, game over — right?  Maybe, but not so fast.

Yes, Perez got a little only over 80% of the two-candidate vote, which makes Imperial very good for him.  But Perez and Yee combined got only 8,200 votes in Imperial County, with Perez having a margin of 4,948.  That means that Perez got about one vote, in a manual recount, for every 1000 votes by which he led Yee.  (We’re no longer worried about percentages of the total vote here; we’re worried about raw figures, with the original magic number being 481.)  So what Imperial is hinting to us is that the “Ballots Required to Add One Favorable Vote,” or BROAFV, conversion factor is about 1000 — a margin of 1000 ballots for Perez over Yee is required for Perez to pick up one vote in the recount.  If so, then Perez might pick up 481 votes through in a population of votes where he led her by 481,000 going into the recount.  That’s the extrapolation to do.

(Ho ho, you might say — there is no such population in this election!  Perez only led Yee by 35,000 in Los Angeles!  Maybe such a wellspring for Perez doesn’t exist — but maybe it’s just hiding.  We’d have to bend the rules to find it.)

Perez’s benefit from Imperial might generate an underestimate of BROAFV if we presume that relatively poor and Latino Imperial was LESS likely than other counties to disqualify questionable ballots.  And, of course, different counties may have different BROAFVs — which is why I don’t think that Perez would stop the recount before he determines what the BROAFV is for Los Angeles.  If it was only 100, for example, then he’d expect to pick up 350 votes even if his recount wasn’t selective!

So, what about Kern?

In the first day of the Kern recount, which was only five precincts, no votes changed for any candidate.  (Perez is reported as having lost one vote, but Parke Skelton reports that they later located it.  I presume that that’s not a joke.)  In Kern Day 2, a count of 24 precincts gave Perez 4 more votes, with no change for Yee.  In Kern Day 3, when 31 precincts were counted, Perez lost another vote.  So that puts him as up 3 votes for 60 precincts.  In Kern Day 4, 33 precincts were counted.  Yee picked up one vote and Perez didn’t change — leaving Perez up only 2 votes in Kern.

We can’t extrapolate from those at to the rest of the 389 precincts that Perez wants to count.  First, he ranked the precincts pretty much from best to worst, so each new batch of precincts, being tilted less in his favor, is less likely to bring him a new vote.  Perez averaged 72 move votes than Yee per precinct over his five best precincts last Friday; for the past three days, that has dropped to 53, 36, and 27 — and, with occasional random exceptions, it will continue to go down from there.

But what we CAN extrapolate from is the margin between Perez and Yee — from which we can assess, as we did for Imperial, how much of an advantage in the electorate Perez seems to need to pick up one vote on Yee. Yesterday, after Day 3 (when 60 counties had been tabulated), the  total Perez advantage was 2,773 and he was up by 3 votes — exactly what the Imperial “1 vote per thousand” rule of thumb would suggest.  After Day 4, his aggregate advantage in the 93 total counted precincts was 3,684 — and he was only up 2.  That’s a BROAFV of 1800 — which is deadly to his chances.  Combine this with Perez getting 5 votes from his 4,948 vote advantage in Imperial and we have him going up 7 votes after counting precincts that had given him an aggregate advantage of 8,632 — or about 1 vote gained for each 1230 vote advantage he has had about precincts.  Wednesday’s count hurt him.  Kern has bad BROAFV.

7. The Benefits of Being Unfair … Uh, I Mean Selective

If we estimate that there may be a 3600-4000 aggregate advantage in the remaining 296 “priority precincts” that Perez has to count, we’d expect him to pick up 3-4 more votes.  That’s only about 10-11 votes — and you can see why people are saying that Perez has to quit.  In a fair count, he’d never make it.

However, Perez has not asked for a fair count.  He is cherry-picking precincts, resulting in an unfair count.  His argument is that if Yee doesn’t like it, she can call for her own recount.  I’m afraid that he is right.

You see, Kern County doesn’t have only 389 precincts — it has 444.  About 20 of those are ghost precincts (or nearly so), so let’s chuck them out.  That means that Betty Yee has about 35 precincts in which she has a slight advantage — probably not enough for her to be expected to pick up even one vote, statistically, but these precincts are important not for what they promise in Kern but for what they portend elsewhere.  These precincts will be counted — they have to be, for Perez to be able to count on any gains from Kern county — but Perez has reserved the right not to hand-count them.  And if they go to a machine count, they will very likely not change at all — perhaps 1 vote in 10,000, and even that randomly.  So in a very real sense, they are effectively excluded from the recount.

In Kern, that doesn’t much matter.  Elsewhere it matters a great deal.  And that’s Perez’s best hope.

8. The Three Orange Counties — and, More Importantly, the Three Los Angeles Counties

Each county in this recount is actually three counties — let’s name them County-P, County-Y, and County-N.  Let’s use Orange County as an example.

OC is conceptually composed of three sorts of precincts: ones that Perez would like to count, Orange-P; ones that Yee would like to count, Orange-Y; and ones that neither would like to count given the costs and time constraints: Orange-N.  These virtual counties don’t have contiguous areas; they’re just formed by three different piles of precinct numbers.

We know that Orange-P consists of 516 precincts, because those are the ones that Perez has asked to hand-count.  If Yee used the same standard for cherry picking counties to be part of Orange Y, she’d end up with 422.  (She won’t do a recount in OC for multiple reasons; primarily, she has better places to deploy her resources.)  The rest would be part of Orange-N — and will almost certainly receive only a machine count.

So the question we should ask ourselves is this: what is Perez’s aggregate margin within only the precincts comprising Orange-P?

Why does that matter?  Because only Orange-P will get the same treatment as Imperial County and the 389 precincts being counted in Kern County.  Orange-Y and Orange-N will get the same treatment as those 35 neglected counties in Kern — a simple, perfunctory, meaningless machine count that won’t do much of anything for anyone.

If you’ve been looking at Orange County and thinking “oh, big deal, Perez only won it by 1.4% and 4,535 votes” — and apparently everyone outside of the Perez campaign has been doing this! — then you’ve badly missed the point.  PEREZ IS NOT DOING A MANUAL COUNT OF ORANGE COUNTY.  HE IS DOING A MANUAL COUNT OF ORANGE-P ONLY.  And within Orange-P he has a significant advantage — twice that of what he had in Imperial County.

In Orange-P, which I define as the 523 counties where Perez won by 8 or more votes, Perez exceeded Yee by 9,054 votes.  Had Perez extended the bounds of Orange-P all the way down to include precincts where he won a by even one vote — we’ll call OC’s precincts with margins between 1 and 7 votes “Orange-Q” — his margin becomes 10,814 votes.

This means that if the “1 vote picked up per 1000 votes of original advantage” rule held up, then Perez should get about 9 votes from Orange County — and, if he wanted to pay for more days of counting, he could pick up an estimated 2 more.  (He has reserved the right to do so.)  If OC’s BROAFV is more like 100 — and I have no reason to presume that it is — then he’d be expected to bring in fro 90 to 108 votes from OC.  That would completely change the ball game.

Orange-P is big.  It provides Perez with a margin almost as large as Yee’s in Sacramento — larger than Sonoma, Marin, Alameda, Santa Cruz, or Monterey provide for Yee, although much less than San Francisco’s.  (And do you know what’s way bigger than even Orange-P?  LOS ANGELES-P!  But let’s stick to Orange County for now.)

What Orange County isn’t, though, is good ground for cherry picking.  Although it has four times the number of precincts as Kern County, OC has only 10 precincts delivering margins of 50 votes or more for either candidate.  Kern has 21 precincts delivering 50 or more votes for Perez alone!  That’s what makes for good cherry-picking.  (It may also suggest larger precinct size up there, but mostly it seems to involve Yee’s anemic numbers in Kern.)

San Bernardino County does not look like it makes for any better cherry picking than OC.  In San Bernardino, Perez stomped Yee by 12,000 votes even before Perez gets to start cherry picking — but there are no areas where Yee’s votes predominated where it would make sense to exclude voters.  (The closest district of any kind for Yee was the city of Upland, which she narrowly lost.)  So I doubt that we’d predict much more than 12 more votes for Perez from San Berdoo, if we presume a BROAFV of 1000.  If there are more, it may be because Latino votes there might be more likely to be challenged than in OC, where it’s harder to get away with it.  And, if you think that San Bernardino has a lot of Latinos and a power structure that is hostile to them, that BROAFV could be far lower once Perez’s lawyers are there poking around.

(Breaking News, 3:30 7/17, and then UPDATE 1, 4:15: Perez has announced that the San Bernardino recount will begin on Monday, apparently running simultaneously with the Kern recount after all.  I’m not going to edit my portion on timing above to reflect this.  And, right after I posted, Parke Skelton tweeted that the two counts will NOT run simultaneously; Perez will stop his manual recount in Kern at the end of the day on Friday, July 18, converting any still uncounted precincts to a machine recount.)

Riverside County also gave Perez a 12,000 vote advantage, but is only a little more fertile ground for Perez’s cherry picking.  The only municipality in which Lee had a non-trivial lead over Perez was Menifee.  So the same analysis applies here to its northern twin.

So, if we presume a BROAFV of 1000, we’re looking at about 24 votes on top of OC’s 11 from the Inland Empire — about 35 in total — and that just used up more than 15% of the state!

An aside: San Diego’s Registrar of Voters doesn’t appear to have a Statement of Votes posted, which is pretty pitiful, but my guess is that not including San Diego County in its list is biggest strategic error that the Perez campaign has made in its recount.  San Diego should allow some very good cherry picking of precincts.  Perez’s team left it off the list, presumably, because Yee outpolled Perez in the county; perhaps they forgot that, if you’re cherry picking precincts, that doesn’t matter.  (I don’t believe that the Perez campaign can add San Diego to their recount now, though, although a third-party could do a separate one — which is what should be done for Yee right now, by the way, if you’re a Yee supporter and this panics you!)

Perez didn’t include San Diego, so they’re still stuck at an estimated 40 votes from the south so far, including Imperial (but not Ventura.)  I’m not going to review the smaller counties to our north — although if even one of them has a lousy and biased ROV that brings its BROAFV way down, it could prove to be a godsend to Perez.

This has all been leading up to the Only County That Really Matters for Perez — Los Angeles.  This is where cherry-picking could pay off big for him — and, if it doesn’t, he’s sunk.

Perez has only asked for manual counting of 700 precincts in Los Angeles out of its total of 4870 — and, unless Perez gives up, I simply do not believe that it will end up being that few.  (In fact, I can’t understand why Los Angeles is a sequential rather than simultaneous count at all, given that a biased and selective count of its precincts is so critical to Perez’s chances.  Maybe excluding San Diego is only their second biggest error in the recount.  Or maybe it’s part of the Rope-a-Dope Perez is playing with Yee.)  With almost 5000 precincts, I just don’t have the time to do the sort of analysis of LA that I’ve done of OC, but I can do a rough approximation by looking at the distribution of votes by community — a mere 279-page PDF that you’ll find linked here.

LA County has already given a 35,000 vote margin to Perez, about 4.8%, so by our BROAFV rule of thumb we’d expect 35 votes from an unbiased count.  That would pretty much finish off Perez.  Luckily for Perez, disparities in communities may give him lots of opportunities to cherry-pick — and even more if the variation within each communities’ precincts is high.

Here’s how many votes Perez extends his margin by simply by excluding all precincts within given communities wholesale.  (This would be a sledgehammer approach, but of course Perez will be able to use a scalpel.)  Communities with narrow margins for Yee aren’t included:

  • Alhambra: 55 votes
  • Claremont: 103
  • Culver City: 411
  • Gardena: 250
  • Hermosa Beach: 90
  • La Canada Flintridge: 112
  • Manhattan Beach: 291
  • Monterey Park: 567
  • Pasadena: 275
  • Palos Verdes (Rancho and Estates) 227
  • Redondo Beach: 208
  • Rolling Hills/Estates: 66
  • Rosemead: 42
  • San Gabriel: 21
  • San Marino: 158
  • Santa Monica: 664
  • South Pasadena: 223
  • Temple City: 98
  • Torrance: 1009
  • Walnut: 71

OK, let me add this up — and remember that we’re looking for a figure in the neighborhood of 400,000 — and I get … 4941,  I left out some cities — like the 9 for Diamond Bar — so let’s just call it 5,000.  Hmmm.  Our rule of thumb says that that garners Perez another 5 votes, leaving him about 400 votes short.  If our BROAFV figure doesn’t change, Perez is finished.  I think that he’ll wait for LA to at least get partway through to find out.

Once again, I’ve left out variation among precincts within the cities, which is another fertile area for cherry-picking.  So, to be fair, I have skimmed through all 130 or so pages of precincts within the City of Los Angeles — and I just don’t see the sort of variation there that would lead to a massive number of precincts for Perez to exclude — even if he wanted to do so.

If dividing each county up into its three parts doesn’t do much in the southern half of the state, adding some more ballots from Fresno, Kings, etc. won’t likely do the job either.  Unless Perez can bring down the BROAFV conversion factor to something like 1 in 300 or ideally 1 in 200, I don’t see where he picks up the votes.

9. Conclusion

When I began this story, I expected to come to a different conclusion.  Maybe additional research would lead me there — but I doubt it.  For now, at least, I don’t think that Betty Yee is in serious danger, but I do think that she has a good opportunity for to travel the state on the recount tour, meeting voters in places where interest will be high due to the local aspect of the statewide story.

I would feel much better, though, if she would at least do her own count of someplace like Mendocino or Sonoma, right now — or maybe but a Sacramento or San Francisco recount on a low simmer — just to let her (and more importantly Perez) know if it does any good.

So that’s how, and arguably how much, a biased count might benefit Perez.  If you found yourself recoiling at the amorality of it all — well, perhaps you might want to contact the Perez campaign and tell them so.

By the way — if Betty Yee wanted to play hardball, she could close this whole thing down next week for probably 10% of what the California Democratic Party just donated to her.  (Reading the Election Code has some real advantages.)  I’ll explain that to the right Yee supporter — either in the Yee campaign or outside of it, but not both — at the right time.  (No charge for anyone who can bring recount money to the table.)

About Greg Diamond

Somewhat verbose attorney, semi-retired due to disability, residing in northwest Brea. Occasionally runs for office against bad people who would otherwise go unopposed. Got 45% of the vote against Bob Huff for State Senate in 2012; Josh Newman then won the seat in 2016. In 2014 became the first attorney to challenge OCDA Tony Rackauckas since 2002; Todd Spitzer then won that seat in 2018. Every time he's run against some rotten incumbent, the *next* person to challenge them wins! He's OK with that. Corrupt party hacks hate him. He's OK with that too. He does advise some local campaigns informally and (so far) without compensation. (If that last bit changes, he will declare the interest.)