Silicon Snake Oil (or how data analytics didn't really win it for anyone)

'Twas ever thus....

Cambridge Analytica is not the first political technology vendor to made big, unproven claims about its abilities. But we live in the age of silicon snake oil. There are millions to be made selling gullible investors and clients on mumbo-jumbo.

And right now it suits the political left - in the USA and the UK - to see that 'silicon snake oil' as some sort of darkly sinister plot to undermine democracy. The truth - despite the ever more scary stuff we see from documentaries like 'The Great Hack' - is that, just as ride-sharing disrupts taxi services and Airbnb hotels, social media disrupts the established norms of political communication, norms that said messages from political campaigns to the public are mediated through TV and newspapers.

As Micah Sifry (from who I've taken the quotation above) reminds us:

When it comes to voters’ decisions about their choice of candidate, most forms of paid political persuasion, including TV ads, online ads, mailers, phone calls, and door-knocking, have no discernible effect in terms of changing people’s minds. That’s the conclusion of a careful meta-review of 49 field experiments all looking at general election campaigns, published by political scientists Joshua Kalla and David Broockman in the American Political Science Review in 2018

We told you so (or rather I was told way back in 1982 before all this Internet malarkey existed) - political campaigning is nearly all about finding people who say they'll support you and getting them to vote. We repeatedly told activists - 'don't waste your time trying to persuade somebody on the doorstep - if they're arguing with you they probably won't be voting for you, get a VI and skedaddle' (or words to that effect). Later on we used profiling and targeting so as to hopefully find more people who say they'll vote for us and spend less time knocking on the doors or ringing the phones of people who won't.

If you're one of those people who think some sort of sinister international cabal is responsible for manipulating thick people into voting for things you don't agree with then you are every bit as thick as those people you consider so stupid they had their minds switched 180 degrees by a Facebook ad. What's happening is that data analytics businesses like Cambridge Analytica - "all hat and no cattle" as one Ted Cruz advisor put it - are trying to exploit the fixed belief that our minds (or rather, other people's minds) are peculiarly persuadable by one particular form of communication, advertising. It's true that advertisers spend a lot of time thinking about what message their consumers want to hear but they also know that, in most markets, the money is best invested in reminding your existing customers what a fantastic choice they made. The same goes for political communications.

There's a quite pervasive idea that political communication should take place in a sort of virtual arena where every voter receives the same information. This, we're told is because targeted communications somehow corrupt democracy by giving Mary a message about animal welfare and Steve a message about hospitals especially if Steve won't see the message Mary gets and vice versa. There's a deal of discussion about 'dog whistles' and hints at sinister subliminal messages. As political parties quickly find out, they are just like consumer brands where sustaining the brand and motivating the consumer is the primary function of communication. What targeting does is to allow the party to talk to Mary about the things that are important to Mary and to connect those interests to the party's programme. But the starting point here is that Mary is more likely than not to vote for that party - if she isn't the communication is mostly wasted.

This is why Cambridge Analytica didn't win anybody anything with its psychological targeting - it's not very effective:

The Cruz campaign stopped using its data. Chris Wilson, the campaign’s director of research, analytics, and digital strategy, discovered that more than half the voters CA had identified as Cruz supporters in Oklahoma actually backed other candidates

So next time you watch a slick TV documentary or read a long 'join-the-dots' screed in The Observer or the New York Times, remember that, for all that CA might be sinister, linked to shadowy dark money, what they sold was the data analytics version of the quack doctor's patent cure all.

....

Targeting works (which is why ASBOs and PSPOs are ineffective, lazy, divisive policing)

"Broken window" theory (or, to use its pompous operational name, "order maintenance policing") is the idea that lies behind a lot of modern policing including, in the UK anti-social behaviour orders (ASBOs) and public space protection orders (PSPOs):

In essence, Kelling and Wilson argued that latent danger loomed everywhere, and everywhere people’s disorderly impulses needed to be repressed, or else. Their “broken windows theory” didn’t stay theoretical: Also known as order maintenance policing, this tactic propelled an entire generation of policing practice that sought to crack down on minor “quality-of-life” infractions as a way to stem violence.

The problem is that, as research now shows, all this sounds good but doesn't work - a great deal of time and money is invested in the idea that anti-social behaviour (essentially annoying but non-criminal acts) and petty crime are gateway drugs to more serious crime and that we should target deprived areas using aggressive approaches like stop-and-search, arrests for petty misdemeanours and large scale area campaigns. This, in most developed countries, inevitably means targeting areas with concentrations of minority residents (black or Hispanic in the USA, black or South Asian in the UK) leading to understandable allegations of racism and high degrees of mistrust in police and criminal justice.

What Stephen Lurie, Alexis Acevedo, & Kyle Ott have shown is that serious crime (they focus on violent crime) is much more concentrated socially than 'broken window theory" allows:

...in over 20 cities, we found that less than 1 percent of a city’s population—the share involved in what we call “street groups” (gangs, sets, and crews)—is generally connected to over 50 percent of the city’s shootings and homicides. We use “group” as a term inclusive of any social network involved in violence, whether they are hierarchical, formal gangs, or loose neighborhood crews. In city after city, the very small number of people involved in these groups consistently perpetrated and were victimized by the most serious violence.

We are talking about really small numbers of people - Lurie describes how:

This held true even in areas considered chronically “dangerous,” like parts of East Baltimore. There, the group member population totaled only three quarters of a percentage point, even as they were connected to 58.43 percent of homicides. Shootings tend to be even more concentrated than homicides. In Minneapolis, we found that 0.15 percent of the population was determined to be involved in groups, but this population was connected to 53.96 percent of shootings—a proportion over 350 times higher than their population representation.

The conclusion here is that we need to do two things: target resources (policing and other interventions) towards the very small numbers responsible for most of the mayhem, and spend some time explaining to the wider public that 99% of young people living in deprived areas with high crime levels are not criminals or even likely to become criminals. Right now with ASBOs, PSPOs and extensions of stop-and-search we are doing the opposite. These strategies act to give the impression to people in these areas that the police and public authorities do not trust them and consider them likely criminals. It is no surprise, then, that these communities - often minority communities - don't co-operate with the police or other public authorities.

We should scrap ASBOs, PSPOs and the whole structure of anti-social behaviour measures and direct resources towards targeting the small number of (mostly) young men who are responsible for most of the violence, burglary, robbery and assaults in our cities.

....

In which the NHS begins to learn about targeting...

****

In a previous life (so to speak) we proposed this to Bradford Health Authority some 21 years ago. Back then targeting was a no-no. It seems the NHS is learning - for example about the impact of alcohol:

Targeting 'risky' male drinkers could save NHS £120m - report

Note that, rather than the usual target,alcohol itself, the proposal is to target the actual problem - the alcohol abuser. And with sensible intervention too:

A 10-minute advisory session with a doctor or nurse can reduce alcohol consumption by up to five units a week.

So much better, fairer and more effective than minimum pricing, bans of advertising and sending out hyperbolic press releases to the Daily Mail!

....

Aim! Fire! - salty thoughts on targeting

Many moons ago I was Account Planning Director at a direct marketing agency. And there we talked about targeting. Obsessively. None of that exploitative ad-gabble about 'brand equity' or 'share of mind' - we talked about targeting. About improving targeting - getting ever closer to the direct marketer's holy grail. To contacting you at the very moment you want to buy what I have to sell you.

At the heart of improved targeting is information - data about you and what you do. There's the obvious stuff - name, address, telephone number and, these days I guess, e-mail. But these are just ways to reach you - on their own these data simply allow me to contact you, they do not allow me to target. If all I have is your address my targeting is determined randomly since I do not know whether you are more or less like to want my goods of services.

So we collected other information and we developed very clever (we thought) systems based on geodemographics (the 'birds of a feather flock together' principle) and psychographics (or 'lifestyle targeting' as the salesman would put it). These systems - built on the back of the electoral register, credit referencing information and other available behaviour data - were aggregated. We didn't know the information about each individual just a set of likelihoods determined by multiple regression analysis. But, couple with a list, we were able to identify the places where the birds who liked our product were flocking, and in doing this to improve our targeting from random.

But this 'profiling' approach - for all its merits and efficiencies - doesn't work that well from the marketers perspective. Despite all the clever number crunching and the melding of more and more information, real behaviour data was always better. Let me explain. Geodemographic profiling will tell me where my customers are concentrated - but it won't tell me enough about my customer's immediate neighbour to generate a sufficient uplift in response.

So why - other than Sunday afternoon indulgence - am I burbling on about targeting? Well, it seems to me that we have to resolve the use of targeting - not by businesses but by public authorities. I recall trying to persuade Bradford Health Authority to use targeting to improve the performance of public health campaigns. Rather than scattering information far and wide in the hope that it reaches the target group, why not use thse geodemographics and other data to get the message directly to the person at risk.

Back then we were sending messages about AIDS to 75 year old grannies and I'm sure not much has changed today. Take salt. For some (but not all) people with hypertension reductions in salt intake are highly recommended as a means to manage heart attack risk. For the rest of us it really doesn't matter - our salt intake in no way constitutes risky behaviour even if it is far above so-called recommended levels. The 20% of the population for whom salt is a risk factor can be easily identified through a simple test (GPs could do this) and the rest of us could go on with having food actually tasting of something.

The same approach could be adopted with other risky behaviours - rather than spending millions sending messages to a general population that isn't at risk (for example in their drinking, smoking or drug habits) we could direct that funding towards those whose risky behaviour does present a problem. We could target but we don't. We could use geodemgraphics, medical records and much else to improve public and primary health but we don't. We opt - for reasons of 'fairness' and equity - to spend the money on general campaigns produced by grand, flash and fancy ad agencies rather than intelligent, targeted direct marketing agencies. And we prefer to ban the agent rather than address the problem user.

Targeting public health campaigns would have a number of beneficial outcomes; Firstly it would mean we get better health outcomes from the spending; secondly, it would get away from the finger-wagging nanny state approach to health campaigns; and third it would allow more risky behaviours to receive public health campaigns. And, of course, it would mean that folk like me who are not massive public health risks and know what we're doing are less pissed off by the hectoring doctors and their chose fake charities.

....