Even if Chris Vance and his fellow Republi-slime weren't working to squash King County's laudable efforts to
produce a maximally-accurate recount and to
correct their systemic errors, the pace of new recount results would be decreasing. Wednesday saw three new county reports, leaving the four largest counties in the state still unaccounted for.
In yesterday's update, I listed the seven then-unreported counties. Below the fold, we'll look at the results with Skagit, Thurston, and Whatcom Counties in the hopper...
Before giving you the new results, I'm happy to note that Cowlitz County changed their report to remove the "Registered Voters" count from the "Ballots Counted" column. Put them down for 42,723 total ballots and 79.2% turnout. I still think there's something odd about their results. The machine recount finished with 99 fewer valid ballots and 69 fewer for-a-candidate votes than the initial count, and the hand recount removed 1 more ballot and another 5 for-a-candidate votes from their totals. I know we won't have so many recounts that they'll end up with all zeroes, but that's the trend.
As stated earlier, three counties reported their results on Wednesday. Here's a summary of their findings:
- Skagit -- losses all around. Gregoire -16, Rossi -5, Bennett -3, and there were 19 fewer valid ballots than in the machine recount. Uses the ES&S Opscan 550 system, about which I'll have more to say in a moment.
- Thurston -- Gregoire gained 15, Rossi added 13, Bennett increased by 1, and the county tallied 2 more valid ballots than in the first recount. Votes with ES&S Punchcard machines.
- Whatcom -- Rossi +22, Gregoire +16, Bennett +2. Counted 14 more valid ballots than in the first recount. Uses the BCCS Punchcard 312 system.
For the day, Gregoire gained 15 additional votes, Rossi added 30, and Bennett had no net change. For Gregoire supporters, these results are a bit disappointing because she outpolled Rossi in two of these three counties. Among them, the three counties tallied 3 fewer ballots than they had counted during the machine recount.
The interim day-end totals showed Rossi with 278 more votes than in the machine recount. Gregoire had added 199 votes to her total, and Bennett had received 7 more than previously (the SoS website still says +8, due to another of those little Cowlitz County glitches). The total ballot tally was 128 higher than in the machine recount. Franklin and Whitman Counties added far more new ballots than the others; in fact, the other 33 reporting counties combined to tally fewer ballots than in the machine recount. The "new-found rate" (additional for-a-candidate votes) continues to drop, to 4.15/10000 ballots. With a net of 484 new-found votes, my estimate of the final total comes to 1195. This obviously doesn't account for the King County/Phillips ballots.
The 35 reporting counties represent just over 40% of the state total ... those four remaining counties are big. Even with the addition of two "Gregoire counties" on Wednesday, only 27% of the recounted ballots come from counties where she outpolled Rossi.
Now back to the ES&S Opscan 550...
Prompted by comments to earlier updates, I did a quick-and-dirty calculation to see where there might be significant differences between overall county results and the additions/subtractions uncovered in the hand recount. For example, let's look at Thurston County, where the major candidates added 28 votes between them. Based on their overall percentages (and pretending that fractional votes could exist), their expected split would be 15.23 for Gregoire, 12.77 for Rossi. Therefore, the actual 15-13 result is very much what would be expected.
Looking at all reporting counties in this manner, I found rather few significant deviations from expectation. In fact, there were only two:
- In Skagit County, Gregoire lost 16 (-9.68 expected) and Rossi lost 5 (-11.32 expected).
- In Grant County, Gregoire lost 5 (expected -0.93) while Rossi gained 2 (expected to lose 2.07).
Both Grant and Skagit use the ES&S Opscan 550. And only one other county in Washington uses that voting machine system -- Cowlitz, where we've seen all those funny events throughout the process. Although the overall candidate effects in the hand recount are insignificant, I suspect I'd find significance if I looked at the machine recount vs. the initial count.
Which is not to say that there's anything untoward going on with the Opscan 550. I report, you decide.