traffic cameras: EXTENDED EDITION


This is the second of my posts about speed cameras. Part 1 is here. Part 3 is here.

I’m amazed by the reaction that my post on ATE cameras elicited. Thanks to everyone who took the time to read and share it. Seeing this level of interest provided inspiration to look at a couple of lingering questions I had (and to learn how to make my MatPlotLib graphs marginally less ugly).

In my last post I showed that ATE speed cameras are highly skewed toward highway traffic. I also argued that looking at the demographics around highway speed camera locations doesn’t tell us much about the equity impact of the citations being issued. That is the main point I wanted to convey and I am not going to qualify it here.

But even though non-highway effects are a less important part of the ATE picture, it is still interesting to examine them from an equity perspective. So I ran the per-ward analysis of camera days and total citations while excluding cameras associated with the largest road classes (“motorway” and “trunk”). By excluding very large roads we can be more confident that the citations are reaching people who live in the area.

Camera days, you will recall, is a metric meant to capture the overall level of ATE monitoring by counting up the number of days that each camera issued at least one citation (indicating that the camera was installed and active).

Maybe ward 7’s a little high; maybe ward 8’s a little low. Nothing seems particularly scandalous, though. Ward 5 looks fishy–could a highway camera be getting mistakenly included?–but visual inspection of these shows nothing amiss:

Ward 5 just has a lot of camera activity along North Cap and South Dakota Ave. Similarly, Ward 7 is showing elevated numbers because of the one RFK site–mentioned in the last post, and notable for generating a ton of citations but not being on a highway–and for having more cameras on peripheral streets (Southern and Eastern Ave) than Ward 8, which is probably appropriate given those streets’ extremely dangerous reputation.

Rereading my post made me realize that the precipitating tweet didn’t describe the DCPC study correctly, and instead referred to red light cameras. My last post focused on cameras that issue speed citations, for reasons I explained at the time. But we can also look at other types of ATE. I came up with the following list of violation codes by eyeballing citation counts. When a code was responsible for 100x as many citations as others, it seemed like a pretty safe bet that it was being generated by robots. Here’s the list I used:


I ran the same query using these codes instead of the speeding violation codes, while still looking only at citations from agencies that reflect ATE activity:

The distribution of these red light and stop sign cameras seems reasonable enough, but it is worth noting that Ward 7 residents are getting a larger share of citations than their share of camera days would suggest (though again: all of this is a small fraction of highway-related citations). Looking at the data, there’s a particular location along 27th St SE that’s responsible for a staggering 77% of the non-speed, non-motorway/trunk citations in the ward–over 117,000 for PASSING STOP SIGN WITHOUT COMING TO A FULL STOP. That one camera is a stone-cold killer; watch out, ward 7.

Lastly, I did want to note one irregularity that’s hard to miss. As discussed already, human-issued citations for speeding are a negligible part of the overall citation picture. Still, I think it’s fair to say that the psychological effect of cops sitting in speed traps looms large. And it is notable that MPD district 7 does basically no human speed enforcement:

That’s right: in the time period studied, district 2 officers wrote more than one hundred times as many citations for speeding as district 7 officers.

(We have eight wards but 7 MPD districts, but it’s not hard to adjust: district 6 serves ward 7 and district 7 serves ward 8, more or less.)

This is something I’d like to dig into more. Did district 7 deprioritize human speed enforcement because of a recognition that ATE does the job better? Because human officers were needed for other, more pressing tasks? Was this a policy change that lined up with the dawning national awareness that traffic stops pose outsize risks to Black Americans? Could it be that ATE’s debut was especially jarring to ward 8 residents because they were accustomed to an era of extremely low speed enforcement that preceded it? I’ll have to load more historical data to answer these questions.

About the author

Tom Lee
By Tom Lee