Crossposted at
Preemptive Karma
This piece was co-researched and written by Carla from Preemptive Karma and TorridJoe from Also Also blog. You'll find it crossposted at AlsoAlso as well.
Since the election contest in Washington appears to be moving forward to the trial phase, we felt it was important to understand the evidence being proffered by the Republicans to have the election set aside.
We want to be clear: King County has made some mistakes in this election. Perhaps the most glaring is the one SoundPolitics has dropped from their radar: the 723 ballots initially rejected for not having signatures on file in the computer when they knew that some signatures weren't yet in the computer. In our opinion, the rejection of valid votes is always worse than the acceptance of invalid votes, unless the result of official fraud.
(more after the jump)
The prime Republican target for allegations of misconduct has been King County. King is the largest county in the state and has become a lightning rod for Rossi, the state Republican Party (led by the obstreperous Chris Vance) and the rightwing talk/blog echo chamber, led on this issue by Stefan Sharkansky and SoundPolitics.com.
But where Democrats in a place like Ohio were frustrated by attempts to gain more information from election officials, we have found election officials in Washington to be generally open, helpful and dedicated. Surpisingly this has remained true even in the face of a vitriolic echo chamber and a heavy burden placed on them by the contest litigants.
These experiences are what gives us such a dim view of the way Sharkansky and SP have gone about their campaign. Too often, the charges have been leveled in a careless manner. Unfortunately, what's emerged is a pattern of discovery and accusation, then publication and THEN perhaps checking to see whether the story's right. Corrections are often made grudgingly and fleetingly.
The most visible mistake was asking the postal service about bulk mailing of military overseas ballots, and not asking about the right bulk mail license. With a minimum of fact checking, Carla was able to suss that one out and defuse the allegations. It's not just election stories, either--this week an associate was incredulous that state laws regarding art set-asides on building projects, would cause an unholy 16mil to be spent on art in front of a sewage plant. Apparently no one asked the development team until later, when it was discovered that the set-aside is not calculated on the total development cost, as SP had blithely assumed. SP corrected their error, but not until well after their article ran.
So when Sharkansky declared with typical hyperbole that he was compiling the "definitive" analysis of King County's votes, we were interested but skeptical.
Sharkansky acquired the public records that were available on the county's results and voter rolls, compared them, and concluded that contrary to King's assertion of around 2,100 anomalies, the total number was actually more than 3,700. Wow! "County Underestimates, Hides Vote Discrepancy 70% Bigger Than Initial Report!" Get Brit Hume on the horn!
Sharkansky's prime contention was that the 2,100 figure represented only a net discrepancy: voters without ballots and separate ballots without voters were allowed to cancel each other out, rather than counting as two separate errors instead of zero. That's a valid charge, if proven. An unmatched vote is an unmatched vote. But the devil is definitely in the details, and that's where it would have helped to pursue some extra knowledge about trying to replicate the county's totals. (Hopefully ignorance is the excuse, anyway).
The two vital pieces in reconciling the books are the voter list, and the tally of cast ballots. Ballot tallies are easy enough to compile, we did it as easily as Sharkansky or anyone else can, by downloading data files available at the County site. King County released a list of registered voters as of Nov 1. The next two releases were lists of voters credited with voting (December 30 and January 7). Both were preliminary but successively refined lists.
Sharkansky's analysis is precinct-based. In other words, he's taken the number of ballots cast in a precinct and tried to match it up to the number of ballots credited for that precinct. If the two didn't match up, he counted the variance. So if he missed one proper registrant, or misplaced one in a precinct when they belonged in, the number would be off.
There are over 2600 individual precincts in King County. Most polling places serve multiple precincts. Cast your ballot at the wrong table and you'll be misfiled. Allow for just one reconciliation error per precinct, and suddenly Sharkansky is vulnerable to 2,600 mistallies in his analysis. King County recorded less than one discrepancy per precinct, according to their figures.
By Sharkansky's own admissions, he struggled to create a file that he believed would match what King worked with to reconcile their data. But his struggle was futile from the start, which he must have known: King didn't reconcile their data at the precinct level, they did it voter by voter, pollbook line by pollbook line. How can you claim you've done the definitive analysis, when you don't even have the right file defined? We don't think you can.
In the next piece, we'll talk more about King's electoral process, and see whether their performance is out of step with the rest of the state.
--TJ and Carla