Photo by Laurence Dutton
Artillery Row

Against the consultancy blob

Why are we still listening to the behavioural insights team?

Could I be a potential murderer? Reading the latest report by the Behavioural Insights Team commissioned by the Mayor of London, there’s every chance I’m exactly the sort of person they’re looking for. Have I ever missed an appointment with the mental health services? Yes, if an aborted course of NHS-prescribed CBT over the phone counts. Do I drink large amounts of alcohol? Sometimes. Do I exhibit “aggressive behaviours” on social media? Well, yes, but only when I read the latest report from the Behavioural Insights Team. 

Understanding homicide: a framework analysis made headlines this last week after its astonishing claim that up 50 recent murders in the capital could have been prevented — presumably at the hands of the recommendations in their report. This is no small feat. Such prevention, if it works, would reduce London’s murder rate by a third at a time when violent crime has been steadily rising for the last six years. It’s no surprise then that Sadiq Khan praised the report as “groundbreaking”, and a key part of his strategy to reduce murder based on the idea that violence is “preventable not inevitable”. Here again was the much vaunted Behavioural Insights Team working their magic. 

This team is not without controversy. Headed by the “nudgy professor” David Halpern, whose big idea is that people’s decision making can be “nudged” towards optimal outcomes by subtle changes in communication and design in their environment. Halpern was a key government advisor during the pandemic, taking a seat at the SAGE table and helping the public make those decisions on everything from social distancing compliance to vaccine uptake. Cue accusations of enabling authoritarianism, though they have since survived this intact, now overseeing a pop psychology empire of consultancy that involves work with some of the country’s leading policymakers and public services. 

NHS services may have already broken the law on protecting people’s data

Beyond accusations of social control, a more obvious criticism is that much of their research is hyper-intellectualised nonsense. As Andrew Orlowski pointed out in 2014 as they were gaining influence in government circles, often their “research” amounted to largely superficial design alterations that brought about marginal improvements to public services. Not much seems to have changed. If you can survive the unbelievably verbose research methodologies, you often arrive at suggestions effectively advocating for things like painting arrows on the floor, or better worded text messages. For all their intellectual clamour — and a curious reputation for attracting the brightest and best — their insight backed solutions all too often hover between mere common sense or a failure to grasp the reality of the service they are advising. 

At first glance, their latest 96 page report on reducing murder in the capital appeared to be a vintage example. There is some wisdom — such as the mapping of murder hotspots (which the police arguably already do). Beyond that, the report is a tough read. If you can survive the explanation of their “comprehensive coding framework” methodology (an extremely roundabout way of saying they looked back at the case notes for the 50 murders and dug a bit deeper), you’ll reach insights that any bloke standing in the street could tell you. Drug dealing can lead to escalated violence; alcohol lowers inhibitions in both victims and perpetrators. Then there is the eye-rollingly nebulous claim that “mental health” is at the root of most murders. In the words of one former senior Met officer I sent the report to: a very clever way of stating the bleeding obvious. 

This brings us to the recommendations, most of which rely on using these apparent insights to trigger an appropriate intervention — usually by social services. Think Minority Report, but Linda from an NHS mental health hub sitting behind a computer looking for patterns of escalation or even … possible mental health problems. Of course, it sounds like a great idea, stopping a murderer in their tracks. In practice, however, having spoken to former Met officers, it’s not entirely clear how the capture of such information would help “target resources”, let alone what an intervention aimed at preventing murder would look like. What does “targeted additional support at people withdrawing from mental health services actually consist of? When would an officer intervene if someone had missed a string of mental health appointments? In a city like London, where nearly 1 in 4 people experience a mental health problem, this sounds like an awful lot of work for an outcome that isn’t clear. 

More importantly, can we really reduce the instance of murder, or even properly understand it by reducing it to a set of often vague social determinants like alcohol use, “mental health and “patterns of escalation?” Curiously, this is something the project lead Ed Bardon perhaps unintentionally acknowledged when he said, “a lot of unusual stuff has to happen for a murder to take place. Despite this, the Behaviour Insights Team seems to believe in the power of data to unravel this mystery, so much so they are apparently keen on building a database of incredibly private information to further analyse this data. The report admits this isn’t easy, but that doesn’t stop it recommending that smaller social services “standardise” their data protection and even set up a culture of “open source data” analysis. 

It’s hard not to read that sentence and wince in light of what’s currently happening in the NHS, where an ongoing attempt to set up a similar database for actionable population health data has seen a whole host of teething problems, some of which may even be illegal. In an extraordinary letter by the Government’s “National Data Guardian” last week, it revealed that some NHS services may have already broken the law when it came to protecting people’s data. Indeed, setting up these sorts of data research environments is no easy business, and it increasingly looks like a monopoly of expertise is set to land controversial US data giant Palantir with a juicy NHS contract. Is it really worth the police flirting with third party organisations to break down barriers to privacy for the sake of providing us with insights that a) we largely already know, and b) we lack police resources to actually do anything about? 

The consultancy blob commands an air of sacred mystery over public policy

There once was a time before the Behavioural Insights Team wrote such reports where murder rates did dip significantly. This is something former Met Borough Commander Kevin Hurley was keen to point out when I asked for his take on the report. According to Kevin, “Operation Blunt 2”, which he helped oversee between 2008 to 2011, more than halved youth murders. What data was used to drive those stop and search decisions, I asked him, fresh from a 96 page Behavioural Insights fugue. “No data, but knowledge of the streets where the officers did their patrols and spoke to members of the local community.” In other words, the knowledge of the beat. You mean the police didn’t search for correlations between different vulnerabilities to build data-led classifications to better understand what was driving violence in the community? “No. With tens of years of experience working with violence and murders, we knew what would work to stop youth homicide. Stop and search, so that’s what we did.”

Regardless of what you think of stop and search, this view from the police — the body mainly responsible for driving down murder in the capital — seems curiously absent. From those I spoke to, the general response to the report was so what? Worse, there is a strong sense of apathy from these former senior officers, who have seen plenty of these reports come and go. At this point, you start to wonder if the writers of these reports would be better off walking the streets of London, or even picking up the Evening Standard. Murder, particularly when related to gangs, is not as esoteric as such a report makes out. There is no need to read an entire report on behavioural insights report on murder in London when it’s clear there is an ongoing problem with young black men being killed in the capital. Or that repeat offenders are more likely to commit violent crime than the general population. Indeed, the data we do collect on the latter group is clearly not enough to stop dangerous and reprehensible individuals reentering society and killing women. 

There is of course a larger problem at play here: the allure of the consultancy blob which has come to command an air of sacred mystery over public policy. Write a report, throw in a few graphics with arrows breaking down the data analysis, add a spice of academic endorsement with something about machine learning insights and voilà. Yet curiously, as the reports get longer, the datasets larger and the experts smarter, our public services seem to get worse and worse. This public service consultancy blob has reigned supreme for nigh on a decade, and I’m not entirely convinced they have much to show for it. Indeed, the pandemic response — that vast bureaucratic failure of public health policy — may well have been its finest hour. This isn’t to say they offer no value, rather that value is somewhat limited, even missing the point, when there aren’t enough people in our public services to do the basics, let alone carry out their recommendations. 

That is assuming those recommendations would even work. I don’t think they will. But by God, I hope I am wrong. I hope the Behavioural Insights Team do cut murder in London by a third. I hope the mentally ill bloke can be stopped from waking up one morning and deciding he wants to stab someone. I hope the woman who has to live with an abusive, alcoholic husband can find refuge, or indeed her neighbours can have the basic civil intuition to suspect something is up. I hope the random act of extreme drunken violence on a Friday night can be stopped in its tracks. Alas, my faith in the Behavioural Insights Team to help prevent those crimes is lacking. Then again, you shouldn’t be listening to a potential criminal like me. 

Enjoying The Critic online? It's even better in print

Try five issues of Britain’s newest magazine for £10

Critic magazine cover