The cliffhanger nears its end. The Washington Supreme Court will soon decide whether those erroneously-rejected ballots in King County should be included in the hand count. While we wait for that, the
recounting continues in King County. Last I heard, they anticipate finishing before Christmas Eve, by which time I expect a decision from the Supreme Court.
This report presents the results from the two counties that reported on Friday -- already previewed in late-day updates in yesterday's diary. I'll also speculate a bit on what it will take from King County to overtake the GOP candidate and win this sucker.
That's what's coming to you below the fold...
Pierce and Spokane Counties completed their hand counts on Friday. In another, admittedly minor, development, the
SoS website rectified the niggling little inconsistency we saw in Cowlitz County, revising Ruth Bennett's machine recount total to 1093 instead of 1094. Bennett's +1 in the hand count now looks right. It won't be quite as easy to explain why Cowlitz's ballot count and for-a-candidate vote totals keep decreasing.
Taking the less interesting outcome first, Spokane County came out with minimal changes to its results. Gregoire added 8 votes while Rossi gained 15 and Bennett added 3 to her total. The county counted 8 more ballots than in the machine recount, which had used the ES&S Opscan 650 and 100 voting machine systems. Spokane's new-found rate was very low, only 1.3/10000.
Although the county was in yesterday's news more because one of its jurists somehow decided to exclude the disputed King County ballots, Pierce County also contributed heavily to tightening the race for governor. A large number of for-a-candidate votes were added in the county's hand tally, with Christine Gregoire gaining 232 votes to Dino Rossi's +201 and Ruth Bennett's +14. That's a 31-vote edge for Gregoire in a county where she'd trailed her opponent by 4%. This result is significantly different from what would have been expected from the prior results. Pierce counted 1 more ballot than in the machine recount. Using the Optech 4C-400 and III Pe optical scanning systems, Pierce found 14.1/10000 new for-a-candidate votes, a rate exceeded by only one other county.
Thus, with 38 of 39 counties reporting, the actual vote counts for the two major candidates are: 1,021,926 votes for Rossi, 867,167 votes for Gregoire. That margin comes to 154,759 votes. At present, our best estimate of King County's outcome is the machine recount, in which Gregoire had a county-only margin of 154,709. Yes, in a sense the gap is currently 50 votes, but in reality that tiny value is the difference between two very large numbers. But now one of those big numbers -- the 38-county gap -- is fixed; the one remaining variable is the King County margin.
The big question, of course, is whether Gregoire's margin in the King County hand count will increase by at least 51 votes. In making educated guesses, we must state our assumptions and then follow their implications. So let me stipulate my basic underlying assumptions:
- All machine-recounted votes for a candidate will again go to that candidate.
- When considering potential additional votes, Bennett will receive the same 2.16% she had previously in King County.
- All vote-count estimates will be rounded to the nearest integer.
Using these assumptions, I set up spreadsheet formulas to estimate the distribution of Gregoire and Rossi votes needed to reach the required 51-vote margin, for a range of new-vote counts.
For example, consider the case where the 723 disputed ballots are allowed into the hand count, all show votes for candidates, and no other new votes are found. In that situation, I estimate 16 Bennett votes; Gregoire would thus have to win 379 of the remaining 707 votes to finish with 1 more vote than Rossi. That's only 52.4% of the total, well below her overall 57.8% pace. In fact, if she maintained her previous pace with 723 new votes, Gregoire's King County margin over Rossi would be 128 and the overall Gregoire margin would be 78.
That's a very longwinded way of saying that if the Washington Supreme Court rules in favor of King County and the Democratic party, Gregoire will win the hand recount.
Another way to look at the situation is to estimate the break-even point. By that, I mean the minimum number of new votes that would be required in order for Gregoire to win while maintaining her overall vote percentage in King County. By my calculations, that point is in the range of 285-290.
With fewer new for-a-candidate votes than that, Gregoire would have to do better in the new ones than she has previously done; with a higher number of new votes, she can still win even if she doesn't do as well in the new votes as before.
Next question -- how many new votes might appear in King County? In this case, we'll assume that the disputed ballots will not be counted.
If King County produces new votes at the rate established by 38 counties (5.86/10000), its total will come to 526 new votes. Gregoire would need to earn only 53.8% of them for the win.
On the other hand, perhaps voting machine technology is an important factor. Three other counties -- Chelan, Klickitat, and San Juan -- use the same Global AccuVote system that King has. Among the three, there was zero net change in the number of for-a-candidate votes. In that unlikely scenario, of course, Rossi holds his 50-vote margin and wins.
As RonK and others have pointed out, King County had already started "divining voter intent" during the machine recount. It's likely, then, that some of what would have been new votes in most other counties have already been accounted for in King County.
In the end, I must repeat what I've said before -- no predictions. Wait-and-see is the only rational attitude.