The NewCo Daily: Today’s Top Stories
The Data Targeters Who Couldn’t Shoot Straight
Since Donald Trump’s upset victory in November, there’s been a steady stream of stories crediting at least some of his win to an obscure data-analysis company called Cambridge Analytica. The company promised to use “psychographics” to target individual voters with digital ads. Its know-how, according to some accounts, helped surface a wave of Trump voters that conventional polls failed to measure.
Trouble is, Cambridge Analytica didn’t actually employ any of its psychographic profiling techniques on Trump’s behalf, and its actual role in the presidential campaign was modest at best (The New York Times). Some coverage has been skeptical from the start.
Here’s how the company describes its “secret sauce”: It takes psychological profiles of a slice of a population, correlates these profiles with other data to develop a predictive algorithm, then, then uses that system to profile individual voters across the whole population. Based on your purchasing data and voting history, CA claims to be able to predict if you’re fearful, depressed, or aggressive. Theoretically, that should let it pick just the right ad message to show up in your feed. Nothing underhanded here, the company says — just “efficient marketing.”
Getting this kind of psychological microtargeting to work still seems to be a ways off. Cambridge did try some of it with the Ted Cruz campaign (its client before it lined up with Trump). But that didn’t work so well: A majority of the Oklahoma voters its program flagged as Cruz supporters actually favored somebody else.
So maybe CA just needs to iterate more. Or maybe it achieved its high profile thanks to connections, not innovations. The firm’s chief funder is Robert Mercer, the conservative billionaire who is also major Trump backer — and until recently, Trump adviser Stephen Bannon served on the firm’s board.
Tackle “Diversity Debt” Early — Or Pay
Debt keeps markets moving along, but too much of it can crush you. Software teams borrowed the concept from finance and dreamed up the idea of “technical debt” — the shortcuts you take in putting off some kinds of necessary-but-tedious work early in a project in order to meet deadlines. Let too much technical debt accumulate and you’re in trouble.
Now Susan Wu (in Backchannel) proposes that we start thinking about diversity in the same way. Startup founders can put off the necessary but sometimes difficult work of insuring diversity in their companies, but over time, too much such “diversity debt” can explode on you. That’s Wu’s interpretation of all the troubles Uber has suffered of late: Think of it as a massive overload of diversity debt coming home to roost.
“Almost every tech company is mired in diversity debt,” Wu writes, “and the larger you get before addressing diversity issues, the more debt you accrue.” For Uber, she recommends “diversity bankruptcy”: A full leadership housecleaning (including the CEO), neutral third-party reviews, new zero-tolerance policies for sexual harassment and discrimination.
But maybe even that isn’t enough: As long as “companies face no real existential threat from bad press about sexism or racism,” the tech industry will not solve its diversity problems (Kate Knibbs in The Ringer). Change this profound never happens just because someone inside, at the top, mandates it; it only takes place thanks to laborious organizing or under the pressure of irresistible external forces.
The Dismal Science Is Not a Science At All
Much depends on the work of economists today, including our assessment of how well we are doing and whether things are “good” or “bad.” So it’s sobering to read a thoughtful essay by Econtalk host Russ Roberts (NewCo Shift) on how fundamentally unreliable the work of economists is. “Most economics claims are really not verifiable or replicable,” Roberts writes. Economists try to create the equivalent of lab studies by using statistical techniques to isolate variables, but that’s very difficult to achieve, Roberts says, “in a way that is reliable or verifiable.” That’s why, when it comes to so many specific issues (like the minimum wage or immigration), “smart people on both sides of the issue each with their own sophisticated analysis to bolster their claim.”
Economists like to say that they don’t rely on theory — they just listen to what the data tells them. But “numbers don’t speak on their own. There are too many of them. We need some kind of theory to help us decide which numbers too listen to. Inevitably, our biases and incentives influence which numbers we think speak the loudest.”
Roberts says the best economists can and should do is to show their work, share their data, and be more forthright and humble about the limitations of their discipline.