Sensible data

By Felix Salmon
January 14, 2014
an essay in the January issue of Wired about the limits of quantification.

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

I have an essay in the January issue of Wired about the limits of quantification. In the magazine it’s headlined “Why Quants Don’t Know Everything”, but online it’s been retitled “Why the Nate Silvers of the World Don’t Know Everything” — which is a little unfortunate, since the whole essay is deeply indebted to Silver’s book, which makes substantially the same point.

The initial idea behind the essay was the concept of priests vs quants — the seemingly eternal (but actually only quite recent) distinction between people who trust numbers, on the one hand, and, on the other, people who rely instead on their personal expertise and experience. Think of the difference between Billy Beane, using dispassionate analysis to outperform in baseball, and Bobby Fischer, whose gifts at chess seemed almost god-given. Nowadays, in a world of quasi-infinite data, it seems the quants are in the ascendant, while the priests are reduced to fighting a rearguard action, clinging desperately to some vestige of relevance. Today, if you want to change someone’s mind, you don’t appeal to authority: instead, you bring numbers.

The result is a deep societal disruption, in which quants take on priests and win: the Oakland A’s against the Yankees, the Obama team against Romney’s. It doesn’t take long before the war is won — we’ve all seen this particular movie before, especially the kind of people who sit on boards of directors. Thus does the priesthood wither away, taking with it a huge amount of valuable institutional knowledge.

The rise of the quants is, unsurprisingly, one of the driving forces behind Silicon Valley venture capital: if you start a small company which competes with a huge company, and the small company gets a few significant wins, then there’s almost no sum the bigger company won’t pay to acquire its smaller foe and use those skills against its competitors. The small company never needs to make money: all it needs to do is show disruptive potential, and it becomes enormously valuable.

But the data-rich narrative — the idea that science is taking over the world — has bred its own counter narrative for some 200 years now, ever since Mary Shelley published Frankenstein in 1818. People don’t like the idea that the computers are in control: for a prime example, look at the way Twitter exploded with privacy concerns as soon as it was announced yesterday that Google was buying Nest.

Or, just look at popular entertainment. The dweeby Q, in Skyfall, tells James Bond that “I can do more damage on my laptop sitting in my pajamas before my first cup of Earl Grey than you can do in a year in the field.” But Q isn’t the hero: Bond is. Similarly, it’s incredibly easy to paint Wall Street quants as the big villains in financial-crisis stories.

When the quants come into an industry and disrupt it, they often don’t know when to stop. They’re young, they’re arrogant, they’re rich and powerful – and they tend not to have decades of institutional knowledge about the field in which they have found themselves. They don’t work for the people who know such things, and they don’t listen to them, either. They’re winners, what do they have to learn from dinosaurs?

Put like that, the risks are obvious. Quants are just as blinkered, in their own way, as the priests they replace – even more so, in fact, since they can be quite Spock-like in their inability to understand the deep role that certain institutional functions are playing. Quants are great at coming up with clever ways of analyzing the world. But that doesn’t mean they’re great at managing institutions, or understanding how their employees might end up gaming the systems that they’ve created.

The solution to such problems is not to disdain the quants, as Bond does Q. Rather, it’s to synthesize the best of both worlds. Look at Southwest Airlines, for example: it has some of the most sophisticated operations geeks in the business, governing everything from fuel-price hedging to the most efficient way to board an airplane. But that doesn’t stop it having a much more human face than its larger competitors. Apple, too, is a prime example: quantitative to its toenails, it nevertheless is clearly governed by overarching principles of human-focused design. And why did Nest succeed where Google Power Meter didn’t? Just because it was designed in a way which made it desirable as a consumer product.

The secret ingredient, I think, is to ensure that managers have a deep understanding of the science being used in the organization — and also of its limitations. Great technology is all well and good, for instance, but if you want it to be broadly adopted, then you need a whole other set of packaging skills as well. And you can’t take technology further than its natural limits, either. It wasn’t really the Gaussian copula function which killed Wall Street, nor was it the quants who wielded it. Rather, it was the quants’ managers — the people whose limited understanding of copula functions and value-at-risk calculations allowed far too much risk to be pushed out into the tails. On Wall Street, just as in the rest of industry, a little bit of common sense can go a very long way.

15 comments

Comments are closed.