Zero the Hero

  If you are the right age, you probably know the tune  My Hero, Zero from watching Saturday morning cartoons. Zero is important for more than just efficient counting; it's useful for posing well-behaved minimization optimization problems (I welcome any mathematics critique). How could we tell if traffic safety efforts were effective? Try to count "lives saved"? People concerned with traffic safety took a long time to reimagine how to think about measuring safety and setting goals. It started in Sweden, of course. With Vision Zero, there's an understandable objective, and, whether talking in absolute or relative terms, shrinking is going in the right direction. Just steer toward zero.

Trying to control a robot in the real world? Then minimize the error between ideal motion and actual movement -- zero error is best. The same goes with any kind of machine learning, minimize the error. Do this by carefully assessing the problem and transforming it into something that can be minimized.

How does this relate to Vermont politics? Politicians often throw around numbers. Usually, bigger is better, and this can mess with how we think about issues. COVID-19 -- still with us -- is a good place to look for political number spinning. 

In Vermont, the message has been "vaccinate, vaccinate, vaccinate." Of course, immunization is very cost-effective and everyone should consider it a civic duty. something that is part of the basic contract of living in the modern world and enjoying the benefits of accumulated millennia of science, engineering, and civilization. 

But how do we measure success? 'Vaccination rate'? The higher the rate, the better, right? A problem with this is that the notion keeps shifting, and that confuses our interpretation of the numbers. 'Vaccination rate' is a homemade concoction. Try looking it up at the World Health Organization, or on PubMed. The term has no basis in epidemiology, and yet, that is all that has been talked about in the Vermont media. 

Vaccination coverage is the widely accepted, clearly defined measure of how well a population has been vaccinated. But even vaccination coverage obscures an important measure, the proportion of a population that is not vaccinated, which ideally is.... zero. Certain medical conditions may prevent the unvaccinated proportion from every reaching zero, but isn't that exactly what we need to know? How many handshakes today are going to be with someone who is unimmunized and unprotected from us? Measuring 'unimmunized' coverage (1-immunization coverage, if I may be so bold as to recast it as a minimization optimization) tells us a lot more about an immunization campaign's performance. Let's do everything we can to ensure that unimmunized is as close to zero as possible.