More numbers released by King County Elections that don't add up. This is yet another indication that the discrepancy of 780 more absentee ballots counted than absentee voters may well be attributable to bogus ballots and also that King County might be cooking the books.

The only thing that can clear this up is the Absentee Ballot Audit Trail that King County was supposed to have created but hasn't released yet.

A lot of numbers follow, but for those who are interested in the details, bear with me --

As I mentioned earlier today, the last "Absentee Ballot Return Statistics" page shows that 566,291 absentee ballots had been returned by Nov. 15 (or Nov. 16, or Nov. 18, it's not clear). This is 2,042 short of the 568,333 reported to the canvassing board in the Mail Ballot Report.

I downloaded the legislative district files from the Absentee Ballot Return Statistics page and cross-checked them against the voter database. The numbers on the web page table don't quite match the numbers in the files. e.g. The 43rd district file contains 41,155 voter ids, not 41,115. More significantly, the 32nd district file is missing, what's there is a file of records from the 31st district. I can work around this.

Of the 531,263 voter ids outside of the 32nd district, I get the following matches:
528,923 voters credited absentee in the voter database;
2,215 voters not credited
91 voters credited provisional
34 voters not found in the voter database

Of the 564,206 voters in the voter database credited with voting, I get the following matches:
528,923 are listed in the new absentee voter id files
34,924 voters credited from the 32nd district not found in the new files
359 voters credited from outside the 32nd district not found in the new files

The fact that there are 359 voters who weren't yet credited suggests that there are really were still some more ballots left to be counted when these files were produced. The outstanding voters appear to have an unusually high number of overseas voters, which would be consistent with the assumption that the uncounted ballots are late arrivals. The proportional amount for the 32nd district would be 20, so let's estimate the outstanding number of soon-to-be-arriving valid ballots at 379.

The total number of rejected ballots as of the first certified count, according to the Mail Ballot Report was 4,111. But this would have included the 566 "Larry Phillips" No-signature ballots which were initially rejected but later accepted. All of these were in the list of voter ids that I downloaded and they were all credited by the time the voter database was generated, so the we can subtract the 566 from the 4,111 to get 3,545 ballots that would ultimately be rejected.

That means that of the 2,042 ballots that would have to be counted (to get from 566,291 to the 568,333 in the Mail Ballot Report) 1,330 would have to be rejected (to get from the 2,215 that were matched to uncredited voters to the 3,545 total rejects) . Not necessarily unreasonable, if you assume that many of the ones that would be counted that late would have been postmarked late.

[UPDATE some of the rest of this analysis is revised slightly due to new information, but the numbers are not affected]

I also gather that the 34 voter ids that couldn't be matched to any voter in the public database represent voters whose ballots were returned as undeliverable, and therefore the voters were dropped from the rolls. Add these to the "rejected" column, increasing the number of rejects by 34 to 2,249, and reducing the number of required rejects to 1,296.

The federal write-in and/or Address Confidentiality Voters (collectively called "service voters") should not have voter ids and therefore should not appear in the list. But they would still have to be counted in the total number of absentee ballots. Thus, there are still 320 more service voters that would have to be included in the gap between voter ids that match the database and the total number of absentee ballots received.

So that leaves 2,042 - 1,296 - 320 = 426 ballots remaining.

A question remains, what about the 91 voters whose ids show they cast absentees, but are credited provisional? One of three things are possible. (1) their absentee ballots were rejected, in which case they probably cast a legitimate provisional ballot. (2) they cast both an absentee ballot that was accepted AND a provisional ballot, or (3) they only cast an absentee ballot but it was credited provisional for some reason. If it's (1), then we need 91 ballots fewer ballots to get to 3,545, so we have 517 ballots remaining. If (2), we have another problem which we don't have the information to address here, but we still have 426 ballots remaining as we do in case (3).

The bottom line is we have somewhere between 426 and 517 absentee ballots that were valid and counted after Nov. 16 but before the first certification, yet only 379 voters to assign them to. Something doesn't add up. And then there are the 226 absentee ballots which somehow materialized in the recounts that don't fit into these numbers either.

That Absentee Ballot Audit Trail would come in real handy right now.