Washblog

As readers of WashBlog will recall, there was an issue last year with some ballot images that were too faint to read.  Looks like they've scrapped that part of the system, thank heavens!

I'm betting that there is no measurable time difference between finding the right ballot image and finding the right paper ballot.  There is a difference in the duplication process--the duplicate info is being entered on a screen rather than being copied by hand, a much faster process.  Thankfully, it is being done by two people, as with the old ballot duplication process.  So the two most critical features of the old system are now standard.

The presorters for ballot duplication are doing a bangup job pulling the ones that require this.  The scanners are counting 100% of most batches.  The largest number of rejects per batch (100-300 ballots) was three.

I asked for data on the number of letters sent out to correct signature issues, the number of returns, and the percentage of those returns that resulted in a validated ballot.  I was told that final results would be published after certification, and that day to day has a lot of variation.  In general, 1-3% of ballots require signature validation.

In addition, they have only started to collect data on the percent of returns that reault in validated ballots.  After the general election there should be better data on this.

I still want mandatory auditing based on the American Statistical Association's 99% confidence level rather than a fixed percentage which may or may not be adequate.  This will have to be a national initiative--I can't see how states can afford to implement it.