San Francisco record set straight

Steven Hill, who participated in the election process in San Francisco, responds to Deryl McCarty's comments on a previous posting.

Deryl McCarty wrote: "Where we disagree results from the San Francisco experience, that showed there will be wild percent swings in the winners' circle candidates percentages driven by small changes in candidate ballot counts at the"bottom" of the candidate list. San Francisco faced that problem during its first run at RCV and their Auditor (or whomever the chief election official in SF is) got a lot of flack for it."

Hill writes: Allow me to correct the record: I am from San Francisco, and assisted the Department of Elections, the Secretary of State, and San Francisco's vendor, ES&S, in implementing IRV in San Francisco. I can tell you categorically there have been NO WILD SWINGS -- zero -- in San Francisco's elections using IRV. In fact, in all races, the candidate initially in front with the most number of first rankings, i.e. the plurality winner, also ended up being the candidate with a majority, i.e. the majority winner.

And in the counting of the ballots, the order of finish has not changed over three election cycles, and at this point over more than a dozen races. I don't know where this rumor about "wild swings" got its life, but there is no truth to it.

McCarty wrote: "So in the next election using RCV he chose not to release any info except how many first place votes each candidate received."

Hill writes: That's not why the Director of Elections chose to wait until Friday (after a Tuesday election) to release round by round vote totals/tabulation results. He did so because in the first election in 2004, when ES&S went to run the tabulation on Wednesday, low and behold, it didn't work! It turns out they had made a very minor programming error that was not caught during the tons and tons of testing conducted by state and federal authorities.

Once discovered, the programming error was easily corrected, because it truly was minor. But it took until late Thursday to figure it out and make the change in a single line of code, get approval from the Secretary of State for that minor change. Then on Friday morning the tabulation was conducted successfully, and results released on Friday afternoon.

But this awful experience left the Director of Elections with a bad taste in his mouth, and so ever since then he has insisted on doing nothing with all the ballot images processed on Tuesday, Wednesday or Thursday, and waiting until Friday to run the first tabulation and release results. We have requested that he start doing it earlier, i.e. on Wednesday or Thursday, but he refuses. It's not because it is a difficult thing to do -- when the ballots are scanned, the ballot images are fed directly into the central tabulator, and it takes literally about three or four keystrokes on the laptop computer to run the tabulation, which takes all of about three minutes to complete.

So it's very easy and straightforward to run this tabulation with the existing ballots that have been processed up until that point. But nevertheless, the Director of Elections refuses to do so. Good luck figuring out what is the proper balance of transparency/openness and releasing of data. But I wanted to make sure that the conversation at least proceeds with accurate information about what has occurred in SanFrancisco.

All best,

Steven Hill
Director, Political Reform Program
New America Foundation
San Francisco, CA