Scheduling Poll Workers

Thad Hall wonders if Poll Workers could be scheduled like Walmart employees. Thad’s ideas are generally fantastic, and I find his out-of-the-box yet highly-informed viewpoint to be refreshing in this field. In this case, though, I don’t think this suggestion would work. It’s all about training and complexity.

Consider what a well-oiled machine Walmart is, that its employees train for days and then work every day refining their training, and that all of their shift workers are redundant from the point of view of customer volume (very few, if any, bottlenecks). By comparison, poll workers train for 2 hours and work maybe once a year, sometimes less: by the time the next election comes around, they’ve forgotten some of the procedures and the rules have changed. To enable a complex scheduling system, poll workers—especially wardens—would require quite a bit more training. Incoming poll workers would have to be briefed on poll site specifics as they come in to work, while people are voting. Crucial bottlenecks would have to be dynamically multiplexed:

  1. For optical scans, the bottlenecks are usually the check-in and check-out desks. There can’t be redundancy here: the only way to speed up that process is to split up the lists logically, say A-M and O-Z (or more). Dynamically splitting the lists up as rush approaches and consolidating them again as rush ends would be fairly complicated.
  2. For computerized voting, the bottleneck is usually the number of voting machines. There isn’t any easy (or secure!) way to scale that up or down dynamically throughout the day.

So, given this complexity and the unclear benefit (how much money would this really save?), I don’t think it makes sense to complicate scheduling. What’s needed for election-day processes is simplicity, as much of it as possible.