Last Friday I reported on some apparent discrepancies between the November 2008 Provisional Ballot Reconciliation Report and underlying voter records.

The Elections staff responded to my questions. Their remarks are detailed and technical. But for the most part it confirms my inference that the ballot accounting and reconciliation was not complete at certification time.

In all fairness, provisional ballot processing and accounting last November was much improved over the horrendous mess of 2004 (which included enough unlawfully counted provisional ballots to have conceivably changed the outcome of the governor's race).

But it is clear that accounting could not have been complete at certification time. Where the report states that 27,482 voters were "Credited (DIMS)", it should really have said "we believe that 27,482 should be credited in DIMS". Not all of those 27,482 voters had DIMS records which could have listed them as credited provisional voters as of certification time. In reference to the statements in King County's response:

1) the "Voter History" table, which Elections staff confirmed to me this week is the definitive list of credited voters, was last updated with provisional voter credits on certification day, November 25 between 3-4pm. Only 27,289 provisional voters are actually credited.

2) The Provisional Voter table shows 27,458 provisional voters, but timestamps show that records for the 166 voters in question weren't fully updated in that table until November 26.

3) The 19 "Other" provisional voters don't seem to be listed as having voted in any DIMS records.

Again, the issue here is not one of improper ballot counting, but that of incomplete accounting and misrepresenting incomplete accounting as complete accounting. Complete accounting is important to help ensure that the proper number of votes are counted in recounts, and that ineligible votes don't slip through and affect a close election. Accurate counts of credited voters is required by law for validation of tax measures. The numbers in the validation summary report are based on voter credits in the Voter History file, which are not entirely accurate.

Misrepresentation is never a good idea. In my opinion, it would have been a better judgment call in this situation to report at certification time that provisional ballots and voters had been reconciled within 200; to explain that it would have been impossible for any races to have been affected by the discrepancy; and to issue a revised report after clean-up. Instead, it looks like the certification report was embellished to make things look better. Never a good idea.

After confirming my interpretation of certain data fields with elections staff, I maintain my earlier analysis: 15 voters were credited with casting two ballots where both were counted, and at least a few dozen provisional ballots were counted in the wrong districts. Elections staff confirmed that their reconciliation efforts are done at the "batch" level and in county-wide summary reports. They do not attempt to reconcile ballots to voters at the district level.

Again, the error rate is much lower than it was in 2004. I commend staff for the improvements that they have made. But there is room for a lot more improvements before the office lives up to Sherril Huff's wildly exaggerated campaign boasts about "Fortune 500 quality ballot accounting" and "Six Sigma" quality control.