Not Quite a Blog

Many of my colleagues were disappointed to hear that Riverside County, California had both not completed its hand tally of paper records but also that this tally lasted well into the new year.

Riverside Registrar of Voters Barbara Dunmore answers recent criticisms in an Op-Ed in the Press Enterprise ("Redundant Tallying"). Among other things, she says:

Of the record 657,000 ballots cast in Riverside County, 72,251 were cast electronically. This was more than were cast any other county in the state. In San Bernardino County, for instance, only 20,000 votes were cast electronically -- a difference of more than 50,000 ballots.

This continues to puzzle me. Most counties severely restricted and discouraged use of DREs in order to avoid having tens of thousands of paper records to manually tally. This almost feels as if Dunmore specifically encouraged their use in order to make a point that the 100% manual count requirement from the SoS would be burdensome for counties that didn't use optical scan technology.

Dunmore goes on to vocalize a frequent refrain:

In the end, the results of our mandated hand tally, once completed, were exactly the same as the results reported by the voting units on Election Day. No discrepancies. No errors. They were 100 percent correct. [...]

While I understand that the secretary of state's requirements are designed to ensure accuracy, Riverside County has utilized electronic voting for more than eight years, and each election's results have proved accurate.

Great! That's exactly what a good audit of working technology should show. Of course, the implication here is "Why should we do this?". Well, we don't look at companies and say, "Jeez, the audits of your books for the last few years have been perfect. Why don't you stop doing audits of your books?" The reality is that audits are an increasingly necessary part of election administration and the post-election canvass process. If anything, we need to regularize more, richer notions of audits -- more than just hand tallies -- into election administration.

Dunmore finishes with a lament about costs of voting systems:

The hand-tally mandate is a deterrent to using the millions of dollars in electronic voting equipment we have sitting idle. It would be a disservice to voters to eliminate early-voting programs because of a duplicative mandate that saps resources. But that is a decision we must carefully weigh in these difficult budget times.

Yes, this is tough; early voting poses a real problem in that the preferred technology, optical scan, is very hard to support for early voting. We want to make early voting easy for voters and election administrators. Unfortunately, the voting systems we have today are both expensive and of low quality.

To be clear, I don't agree that a 100% count is required, technically, of each vote cast on a DRE+VVPAT system. In fact, manual tallies of a random sample of machines where the sample size is contingent upon the margin in close races should do the trick nicely. Can anyone defend a 100% manual tally of DREs? To me, it only makes political sense as a deterrent.

We want smarter audits, not blunt ones.