Elections as measurements

 

The complete debacle in the recent (Nov. 8^th, 2022) election for the Oakland School Board (OSB) led me to think more about elections.  For those of you who are from outside the San Francisco Bay Area, after tabulating and announcing the election results, the Alameda County registrar found an error in how ballots were counting, and On Dec. 28^th announced a new top vote-getter, less than two weeks before inauguration day.

 

Oakland is a city of about 440,000 people just south of Berkeley (where I live).  The OSB is important, since Oakland schools are facing many problems, including declining enrollment, educational recovery from Covid closures, and financial problems.  The OSB election used rank choice voting, so counting took time; the results were announced by early December.  Nick Resnick won in District 5.  This result was duly certified.  

 

Then, on December 28^th, the registrar dropped a bombshell.  There was a mistake in tabulating ranked choice votes, and a different candidate, Mike Hutchinson, was the actual top vote getter.  Some voters did not select a first-choice candidate, but did select ones for second or third choice.  Those second or third choices should have been tabulated when the ranked choice algorithm got to the second or third choice, but they mistakenly were not counted at all.  This was pointed out by a non-profit that looked over the voting results and spotted a problem.  When this was fixed, the District 5 results changed.

 

Despite this bombshell, the originally-certified winner, Nick Resnick was sworn in to office on Monday (Jan. 9^th).  The ultimate disposition is in the hands of the courts, since Hutchinson has, not surprisingly, sued.   The first hearing will be in May, after many OSB meetings and votes.

 

What does this have to do with physics?  Elections can be considered measurements of the will of the people.   Some measurements give very clear results, while others are ambiguous.  Like measurements, elections have statistical and systematic uncertainties.   

 

The statistical uncertainties are from random fluctuations in who did or did not vote. If the election were rerun, some different people would vote, because they were sick that day, or out of town, or just forgot.  Further fluctuations come from the lag in voting when people move into or out of the district, turn 18, or pass away. Mail-in ballots change the details, but not the overall picture – fluctuations remain.

 

From the measurement analogy, the statistical fluctuations are roughly the square root of the number of voters.  If 100,000 people cast votes then the uncertainty is the square root of 100,000, or 316 votes.   There are other, more sophisticated (binomial) formulae, but this is a reasonable estimate.   Elections with a smaller vote difference could easily have gone the other way. 

 

Systematic errors may be larger, and stem from systemic issues, such as the OSB debacle.  One difficulty with the analogy is that not everybody agrees on what ‘features’ of our election system are systematic uncertainties and which are ‘the rules.’     Funding inequity and unwarranted voter suppression are both major issues which can lead to unfairness, and incumbents certainly have advantages.  It is a matter of defining the question we ask.  Are we measuring ‘the peoples will, according to rules, or ‘the peoples will, as would be measured in a perfect, unbiased system?  The problem with the latter choice is that people disagree about the biases.  ‘According to the rules’ is a clearer baseline. Methodologies exist to try to estimate biases for the latter case.

 

Either way, though, when elections are within the combined error, the vote counts are statistically indistinguishable.  This is not to say that the different candidates are similar, or that partisans on both sides will not feel strongly about the result.  But, there is no discernible difference in the people’s preferences.

 

Treating elections as experimental measurements can help put their results in perspective.   a 51%:49% split is not a mandate, but a small, and perhaps statistically insignificant difference.    Successful candidates will govern better if they keep this in mind.

 

The opinions expressed here are my own, and not necessarily those of my employer, colleagues, family, friends, or anyone else (although they should be).