Guinea pigs for an unproven technology
Quick notes from the ShotSpotter Connect subcommittee meeting
Some quick notes on the 2.5 hour and ultimately fruitless public safety subcommittee meeting last night about the ShotSpotter Connect program, which if you don’t know is a proposed $200,000 investment into the use of artificial intelligence to “forecast crime.” Read some of my past work here and here and here for background and if you like it maybe subscribe? Jk. Unless….
The clear and most important takeaways are that Councilors Kate Toomey and Donna Colorio are obvious “yes” votes, and Councilor Sarai Rivera is on the fence. As this heads to the full City Council, support for the program is similarly mixed, leaving a slim path to victory. This could get voted down and if it does, it will be by a hair.
While she didn’t outright criticize the program in any substantial way, Rivera did call for more outreach to groups which represent people of color in the city. She wants to see meetings between representatives of ShotSpotter and those community groups before she votes on it. Those meetings will likely consist of the same marketing pitch that ShotSpotter representatives gave the public safety subcommittee tonight, a pitch which, as many callers pointed out, is rife with overly emotional language, salesmanship, and a few obvious and glaring factual errors. It is not, by any means, an objective evaluation of the program. It is a marketing pitch from a company trying to convince a municipality to buy their product. This marketing pitch was not counteracted by invitation of any experts in machine learning in general, predictive policing, or racism in police data. We were expected to consider this marketing pitch a serious intellectual evaluation of the program. It was not. In fact, the program is so new that there has been no serious evaluation. One of the ShotSpotter representatives outright said that, a fact which should be worrying to any responsible elected official regardless of their feelings about the police.
In fact this program has only been deployed in some seven other cities. Later in the night a caller said that deploying this technology would make Worcester residents guinea pigs for an unproven technology and I think that’s about as good a line as any to break through to people who don’t so much care about the racism baked into policing patterns and the data generated by said patterns. Why are we rushing into this? Kevin Ksen, a longtime follower of the City Council, remarked he’s never seen anything like this.
The presentation by the ShotSpotter people was long and wordy and glossy but the gist of it was don’t worry folks, this predictive policing program isn’t as racist as all the other predictive policing programs because we only rely on geographic and census data, we don’t explicitly use race data. But in a city like Worcester — pretty much every city in fact — geographic data and race data and socioeconomic data are the same data. There are neighborhoods that are more heavily policed than others and those neighborhoods are poorer and more diverse. It is not a hard calculus and it is reflected by years and years of crime data here in Worcester. That crime data is what the algorithm of this new program will use to predict where crime will happen next and surprise surprise it’s going to be the neighborhoods that already feel the full weight of the boot on its neck. It will only have the effect of further concentrating police in the neighborhoods the police oppress. So that line about Shotspotter being less racist than the other guys is flatly a load of crap and, like they themselves even said, there’s no way to prove otherwise.
An academic with an intimate understanding of algorithms called in and asked some serious questions about how the ShotSpotter algorithm works. The questions, about what sort of algorithm it is and whether there will be an open source component, were ignored by Toomey. Another academic followed up, asking Toomey why she did not refer the questions to the ShotSpotter representatives on the line, and Toomey said she was going to pass them in writing to the company and they will then provide a report back to the council some other time. What could have been the most substantive and intellectually rigorous portion of the meeting was effectively shot down by an elected official who, as part of her brilliant cross-examination, asked how the police department currently deploys patrols.
I so wish I was kidding. This is the chairwoman of the public safety subcommittee, the body that is supposed to examine police policies and procedures in a more in-depth way than the full council would, and we have the chair of that board asking the police chief how they send patrols out. Feels bad man.
Should the program pass, the police and ShotSpotter representatives promised to share quarterly reports with the City Council. I’m just putting this here for posterity when that inevitably doesn’t happen.
While there were many callers who spoke out against this program, there were some who spoke in favor. And the comments ranged from “let’s just kill all the bad guys” type stuff straight out of a Paul Verhoeven movie to “why don’t we just let the police decide what they want” type of stuff which is… straight out of a Paul Verhoeven movie. Friendly reminder that police officers are actively sending emails out to all the neighborhood watch type groups to have them call in to support the program with some lazy talking points which I’ve covered before but here they are again. I’ve got my hands on emails showing at least three officers doing this.
A brief description on what ShotSpotter does:
Detect gunshots for consistent, rapid, precise police response
Detect highest crime risk areas for directed patrols to maximum crime deterrence
Protect your officers with enhanced situational awareness
Protect your community by disrupting crime cycles with data-driven solutions that limit bias
Connect with your community and build trust
Connect to data and insights to drive impactful and measurable outcomes
It’s nonsense words and it’s laughable next to Defund WPD’s rigorous interrogation of the program.
As Defund WPD pointed out in a statement released after the meeting, critics of the program outnumbered proponents three-to-one in the public comment section. They made a bunch of great points, like why can’t we spend this money elsewhere, what is the rush, how do we know this is going to do anything at all besides be actively harmful, and why is the money coming from the City Manager’s contingency fund, a fund that is supposed to be for emergencies?
Also, Defund WPD listed the cities in which predictive policing programs like ShotSpotter have been cancelled after community pushback. And more importantly for the elected officials making the decision to buy this product, cities with open lawsuits for the use of predictive policing programs.
Much to think about.
The subcommittee let everyone, or more or less everyone, speak, but I don’t know if they were really listening. The fact is they don’t really have to. There is only one out of eleven councilors on the City Council who is willing in any significant way to speak out against the police (Khrystian King). Without more significant political leverage, large scale showings of disapproval like Defund WPD has been doing for months won’t really go anywhere. The council can just dismiss the criticism. If we had one or two or three more councilors that were willing to seriously examine police programs, policies, and procedures, we could get somewhere. We don’t have that now, but we do have a municipal election coming up in November, and a couple good candidates could go a long way. There are a couple already in the running, and I’ll get to that in a later post, but it could be you!
Today is the first day you can take out nomination papers to run for office. Just something to think about!
The ShotSpotter program is going to go back to the full City Council for a vote before it is approved and there still have to be various community meetings as decided on yesterday by the subcommittee. So it’s not getting approved tomorrow. There’s still time to try to get a few more councilors on the right side and get a rare victory here.
Let’s go!!
~/~
I know this is just a little guy compared to how long I usually go but I don’t want to completely beat people to death with this ShotSpotter stuff. If you liked what you read please consider subscribing so I can have more and more time to do this stuff or sharing it so more people read.
Tonight’s the full City Council, and the big thing to watch out for is the discussion around Coes Pond and Hillside Beach. I won’t be able to tune in tonight as I am recording a podcast episode, more on that later :-), but let’s just say you should expect something from me on the Hillside Beach situation very soon.
That’s all for now toodeloo!
I still don't understand the appeal of Kate Toomey as a representative. That's not true, I get it. She's personable and plays nice with the entrenched power structure. If you like your representatives to be competent, however...