What to "Expect" from the Remaining Ballots in Seattle

Thursday and Friday I had a friendly online discussion about the Mayoral race with Sandeep Kaushik, a leading voice for Seattle’s centrists and moderates.

Sandeep wanted to make it clear that the comfort I was offering my progressive friends shouldn’t be taken to say the outcome was clear cut.

The gist our discussion hinged around a few issues: I had staked out the position that if historical trends followed anything like their normal course, Katie would win. Sandeep didn’t disagree—but he did note some reasons to doubt they would follow their normal course. These included the weirdly rough count for Katie on Wednesday and the lesser relative improvement she saw compared to other progressive candidates on Wednesday and Thursday. He thought these could be grounds for expecting a progressive shift that is outside of recent historical precedent.

I agree that these form a reasonable basis for doubt. But I also don’t think they win the day. I noted four reasons that I on balance believed the norm would hold:

1.  Common things are common. That should be a very sticky starting assumption.

2. The current shifts were well within normal range.

3. If there is a progressive-shift penalty, it hasn’t been big enough to be decisive.

4. KC Elections was behind on counts; peak shift happens late. Penalty or not, better counts are (likely) coming.

In any case, Sandeep wasn’t saying he was sure we will depart from historical precedent and I’m not saying I’m certain we will keep to it. Our only substantive disagreement here was probably our difference in estimation of whether this will be the case. I’m guessing he’d have said that’s a toss up. I’d say it is more probable than not that we stick inside historical precedent.

After Fridays’ drop, I maintain that position. Katie’s count moved to within striking distance, yielding a daily drop rate (54.84%) that, if continued, would result in a razor thin progressive win.

So the remaining question is—what to expect from remaining votes? Will they come in at just under 55% or better? Or not?

While no one can be certain—history does still offer us a guide. Mondays are usually worse than Fridays (which had me worried for a bit). But fortunately, I was able to work with Daniel Jones again to find the best way to understand relevant history in Seattle. On Friday afternoon, it appeared that 17% of the vote remained. So Daniel looked at past votes to figure out how the daily vote share trends tend to continue.

When I looked at Daniel’s graph, it thought it looked to me like nineteen of his twenty-one examples show increases in (daily) vote share for the last 17% of votes. It looked like one was was flat (which is also okay), and one was slightly down (risky). But I wasn’t sure how to read this precisely, since the last few votes are so volatile.

So Daniel kindly came up with a clearer way of showing the data. Of the 21 examples he found, the last 17% of votes always had a slightly-better-to-significantly-better share of votes for the progressive, with an average of a 2.75%. These range from very small (Deborah Juarez in 2019, +0.12%), to pretty significant (+8.07% from some guy named Ron Davis in 2023—clearly the magnitude of vote shift doesn’t translate into winning!).

So if Katie Wilson’s remaining vote share follows this pattern, she wins.

Are there reasons to caveat this? Yes, of course. The issues Sandeep raised are still relevant. We could break with recent history. That usually does not happen, but it still happens a fair bit. Also, these 17% increments are estimates based on a data scientist’s extrapolation from historical ballot drops. The technique is probably sound, but it could be off by small increments, meaning historic precedent may not be quite 100% in our favor.

Still, the overall picture looks pretty darn favorable to me.

Historical precedent leans overwhelmingly toward a win. I’ve seen some reports of slightly higher turnout data (more wiggle room for a slightly lower vote share, and more likely to mean a bigger increase in daily vote share). I also saw one unofficial report that these later added voters are disproportionately young. That’s amazing!

Also, as I can attest as a volunteer, the Wilson volunteer machine is rapidly curing ballots with great success. Spoiled ballots trend toward the young, and so cured ballots give a good boost to progressives vote shares.

I’d say that, overall, I agree with prediction markets—both polymarket and predictit have Katie at 70% and Bruce at 30%.

Don’t pop the champagne yet. That’s barely better than 2 to 1 odds.

May the odds be always in our favor.

(Apologies for more typos in these last few posts—I’ve been needing to get them out in narrow windows of time, so I haven’t been able to lean on my editor friend for help like usual).