The Resistance Kitchen had a short, but insightful exchange, on a public forum, with Lindsey Chiswick, the Director of Intelligence for the Metropolitan Police Service, who is also the National Police Chiefs’ Council Lead on Facial Recognition.
Interview
Policing Insight online magazine is a policing insiders resource published by CoPaCC – an advisory service for police and criminal justice organisations. They have privileged access to the top police brass which community organisations unfortunately are not permitted. Their interviews and articles give insights to police thinking that are often missing in the public sphere. Although access to most of the resources is restricted to those who can afford hefty subscription fees – usually police forces, a few are available free upon registration. One such article was an interview with Lindsey Chiswick.
1 in 6,000
The tempo of her interview can be gauged by the title quote “Lindsey Chiswick: ‘FR research suggests we could expect a false alert rate of 1 in 6,000 – I’m seeing nothing like that’”, a reference to the bogus Met claim that the error rate in their live facial recognition (LFR) is only 0.017% (1 in 6,000) – in the interview Lindsey Chiswick seems to claim its even less than that! In reality independent studies based on the police’s own data accumulated over 8 years (2016-2023) shows that 85% of the time the system gets it wrong with an innocent person, not on any police watchlist, accused of being a criminal (Met deployment records via FOI, BBW, May 2023).
The whole interview seems to follows this type of delusional belief which seems prevalent in the Met.
Biometric Checkpoints
LFR deployments have been described as biometric checkpoints, with police vans armed with facial recognition cameras scanning each persons face, capturing their biometrics without their consent, and putting them in a police line up against a watchlist of wanted people. At the top of the article in a quote box Lindsey Chiswick is quoted extolling the virtues of these checkpoints: “The benefit is that LFR precisely picks out the wanted individual in the crowd. And the real plus point – which I think we sometimes miss and certainly the media does – is that all those other people are allowed to go peacefully on their way”. As one commentator put it “Pray tell what would have happened to all those people if FR wasn’t being deployed. Would they still not be allowed to go peacefully on their on way?”
Our Exchange
In the interests of community and fairness, we have reproduced the interview below (fair use policy), acknowledging that the copyright holder is Policing Insight.
Original Article (available free upon registration): https://policinginsight.com/feature/interview/lindsey-chiswick-fr-research-suggests-we-could-expect-a-false-alert-rate-of-1-in-6000-im-seeing-nothing-like-that/
It was published on 30th July. We responded with a comment the next day on 31st July which prompted Lindsey Chiswick to respond the following day.
Despite it being a very short exchange it’s still very instructive in how the Met views things, in particular how expendable the lives of people of colour are.
We would really appreciate a proper discussion with the Met about facial recognition as promised by Lindsey Chiswick.
Police Justification
One of the things that came out of this interview and a common justification used by many senior police officers is that they claim that the reason they are inflicting LFR surveillance on our community is because the community has asked them to do it, and a large section supports it.
The reasons for this are three fold.
Firstly, they use fear to cower the community into submitting to such dystopian surveillance – threats of knife crime and other violent crime running rampant without LFR. This is despite the fact that the police’s own research studies have consistently shown that CCTV cameras have zero impact on violent crime and does not reduce crime in city centres (car parks is one of the only places where they are shown to be a deterrence).
Secondly, community organisations whose survival depends on funding from the Met, often directly but sometimes via MOPAC (Mayor’s Office for Policing and Crime), are leveraged to rubber stamp police policy and smooth over police blunders in the community.
Thirdly, perhaps most importantly, it’s through police misinformation – like the 1 in 6,000 statistic – people in the community are bamboozled in to supporting something which in reality targets them.
To counter this false narrative, a wide cross-section of the community has come together to support a letter which we will be sending a to the Met this week. It outlines the reasons for the community’s opposition to the police’s facial recognition surveillance of our community. We hope it will initiate the conversation we were promised.
Additional Question - Will You Obey The Law?
An additional question we asked Lindsey Chiswick was regarding whether she would confirm if the Met would obey the law at LFR deployments.
People are legally allowed to cover their faces when walking past a LFR van unless a Section 60 order (of the Criminal Justice and Public Order Act 1994) is in place – this is rare. This needs an inspector to authorise, giving justification, is restricted to a specific location, and lasts up to 24 hours. During which time police can carry out suspicion-less stop and searches and and demand you remove your facial covering.
But in reality, in the past police at deployments have forced innocent people against their will, and against the law, to uncover and submit to being scanned. Even fining them £90 for complaining.
Again she unfortunately did not reply.
An interesting side product of the exchange was a reply from another reader supporting the Met, and wishing to clarify for us how false positives should be measured with a couple of links to his website explaining the process (the website is linked to his account so we are assuming its his company). At the time I didn’t want to side track the debate away from Lindsey Chiswick, so after a quick look, I just politely replied his company’s method was different to the the way the Met was calculating false negatives and left it at that.
An interesting issue was raised which its useful for us to look at to understand how statistics are used to hide the truth. The Met is not unique in trying to disguise bad results by disingenuous statistics. The linked company is a security consultancy that provide facial recognition surveillance solutions to “law-enforcement, intelligence, defense and government agencies”. So despite their slick marketing banner reading “Ethically Enabling a Safe and Secure Society”, the bottom line is it’s in their interest to show good levels of accuracy to their potential customers. So unsurprisingly they chose a method to calculate false positives which is even more outrageous than the one the Met uses.
The Met, instead of dividing the false alerts by the total alerts to determine the error, divide instead by the estimated number of people passing the van – so their figures can be out by a factor 100,000 if they estimate a 100,000 people walked past the van (note that they don’t actually know how many people walked past – its just an estimate they make to help their stats). This company in their statistical manipulation, have decided to first multiply the estimated number of people walking past the van by the size of the watchlist (ie the number of estimated comparisons made against the database) before dividing. So when the Met are out by a factor of 100,000, this company for a watchlist of 10,000 will be out by a factor of 100,000 x 10,000 = 1,000,000,000, yes a billion!
So imagine, for the same footfall and watchlist size, if their system stops a 1000 people, and every single one of them is an error – all innocent people not on any watchlist. Their stats will still report an error of just 1000/ (100,000 × 10,000) = 0.0001% ie 1 in a million! Must be great for racking in profits from gullible clients.