Could We Apply Trump’s Big Data Tactics… for Good?
If you’re the type of person who’s reading this, you’ve probably already read the Das Magazin/Vice story about how psychological profiling data (called “psychometrics”) enabled the Trump campaign — and the Ted Cruz and Leave.EU/Brexit campaigns before it — to microtarget voters at a level of precision and scale never before seen.
In case you haven’t, here’s the short (and admittedly unscientific) version: everyone’s personality can be broken down into five component parts. We can all therefore be categorized based on where we fall on those five spectrums:
* Openness (how open you are to new experiences?),
* Conscientiousness (how much of a perfectionist are you?),
* Extroversion (how sociable are you?),
* Agreeableness (how considerate and cooperative are you?), and
* Neuroticism (are you easily upset?).
It’s called the OCEAN test, for the acronym derived from the first letter of each of those words. Some call it CANOE. Same letters. It’s not new — it’s been around as a psychological theory for decades. What is new is how social media and big data techniques enabled it to be employed in the two most earth-shattering elections of recent times: Donald Trump’s surprise election victory in the United States and the Leave campaign’s surprise victory in the UK’s “Brexit” referendum earlier that year.
A company called Cambridge Analytica, which counts Trump uber-advisor Steve Bannon as one of its Board Members, has used all the data sets it can get its hands on to develop personality profiles of all 220 million adults in the United States. They say unabashedly that they have amassed 3,000–5,000 data points on each and every one of us. With just 70 Facebook likes, one of these models can predict a respondent’s answers to questions better than the person’s friends. With 300 likes, the model is more accurate than the person’s life partner. Imagine what they can do with 3,000 data points, including not just Facebook likes but also television watching habits (all your Netflix and digital cable choices are recorded in databases, available for purchase), buying habits (do you have a supermarket loyalty card?), credit reports (Experian), and publicly available data sets like land title registration and voting records. When you know someone that well, it becomes increasingly easy to push their buttons… to influence them.
For the small fee of about $15 million, Cambridge Analytica used its data to hyper-target messages to key voters in the US elections, aiming to energize potential Trump supporters and de-motivate potential Clinton supporters in swing states.
Critically, they focused their messaging on social media and digital television, where individual viewers can be identified and shown different versions of messages based on their psychological profiles. This shift in how we consume news and entertainment— from mass media to individually tailored media on our computers and web-enabled TVs — made it possible for Trump to deliver individually tailored messages to the voters he needed the most, and could influence the most easily.
For example, while working for the Ted Cruz campaign ahead of the Iowa caucuses, Cambridge Analytica was able to identify a few thousand potential Republican supporters who needed a little persuasion to head out to their caucus, were interested in gun rights, and tended to score low on openness and neuroticism. They could then nuance their messages to those people to communicate in ways they would be mostly likely to respond to.
When messaging potential voters about gun-related issues, a person who scored high on neuroticism could be told that a gun is a good way to protect themselves from potential burglars, while someone focused on family and tradition would be shown a father and son hunting together at dusk. Different people, watching the same TV show, getting vastly different messages, from the same candidate.
On one day in October, the Trump campaign apparently sent out over 175,000 different Facebook ads, each specifically targeted to its recipient, down to the titles, colors, captions, and videos used.
It’s hard to prove how effective these techniques may or may not have been, but all three campaigns the company applied its methods to in 2016 clearly did much better than expected.
As this #bigdata story continues to spread, many will be tempted to ask “how can we stop this from ever happening again?” The truth is, we can’t. Psychometric theory has been around for decades. The commercial availability of billions of highly accurate data points, combined with the ability to hyper-target messages through digital TV and the internet, have just turned psychometric theory into an advertising reality. Donald Trump seems to have stumbled across it at the moment his campaign really needed it, squeezed just enough value out of the technique to win himself the presidency, and by so doing, brought it to the popular consciousness. For better or worse, psychometrics will be a part of our reality for the foreseeable future.
We definitely need to ramp up data privacy and civic education efforts, to arm the average citizen against those who would seek to influence us too easily.
But another question we need to be asking in parallel is, can we put some of these approaches to use to improve lives — even in small ways — to build tolerance and peace and prosperity for all people?
Could social workers use psychographic messaging to help alcoholics avoid that next drink, or to encourage abusive parents to contain their rage before it gets out of hand? Or to convince vulnerable youth to take measures to avoid pregnancy? Could teachers better use psychographic profiling to tailor lessons to students’ ways of learning? How could we equip them to do this?
Could local governments use these techniques to help young families recognize the value in careful financial management and saving for the future?
Farther afield, could these techniques revolutionize the behavior-change communications work development organizations have been doing for decades in Africa, Asia, and Latin America? Could we tailor messages much better to rural women about sleeping under mosquito nets, or traveling to a clinic to give birth, for example, if we understood their psychological profiles? Could mediators and social justice groups use these methods to promote peace and understanding between different ethnic and religious groups?
Cambridge Analytica succeeded because they could get their hands on 3,000–5,000 data points for every adult in America — most of it based on our everyday life choices, and thus highly accurate reflections of the real personalities of each individual concerned. While that amount of high-quality data isn’t yet available in developing countries, might it soon be?
One indication that it might be comes — worryingly — from inside the corporate headquarters of Cambridge Analytica itself. The company appears to be looking to hire Senior Political Strategists with the “ability and temperament to…[lead] multi-cultural teams working in developing or transition countries...[and an] understanding and experience of electoral politics (particularly in a developing country context).”
The company that probably elected Trump seems to be gearing up to deploy its methods in some of the world’s most fragile democracies. As if they didn’t have enough electoral problems without Steve Bannon’s data scientists profiling voters for the highest bidders.
It’s clear that, from now on, big data and psychometrics are going to be used to target messages to people vulnerable to persuasion, all over the world. It’s critical that those on the side of peace, justice, and tolerance learn to make use of these techniques, because those on the other side are already moving ahead.
Please leave your comments on any aspect of this that rings true/false/important to you — let’s start a discussion about how to use psychometrics for good… if that’s even possible.
