Skip to main content
Fascinating Blog

The limits of data-driven punditry

By July 6, 2015December 20th, 2015No Comments

The progressive website “In These Times” has a interesting … by which I mean “odd” … critique of centrist pundits:  they’re too data driven.

“But at some point, Klein and company stopped being liberals,” Bhaskar Sunkara writes.  “They even stopped being human. The singularity—a technological superintelligence—was upon us. The wonks had become robots, ready to force enlightenment down our partisan throats.”

Wait.  What?  We don’t like facts?  We don’t want to be post-ideological?  One of the wisest things I ever read came from Jaques Barzun:  “An ideology is a philosophy that has stopped accepting new evidence.”

But there’s a point here, to the idea that “Beltway liberals today prefer to tout their expertise and talk raw facts. It’s what distinguishes them, in their minds, from the “ideologues” (people with a coherent worldview) to their left and right. Moral and ethical appeals to voters are thus discounted.”

I remember I was in a Vietnamese chicken joint in San Francisco (if you know the one, you know …) in leading up to the 2008 election.  I was eating alone, minding my own business, but the talk around the counter was all about Obama.

“Thank God we’re finally going to have a president who’s smart,” said a man who obviously worked in software development and almost certainly didn’t believe in God.

“Yeah,” said his friend, who obviously worked in online marketing.  “I can’t believe how much stupidity we’ve had to endure the last eight years.”

“Obama’s brilliant, though.”

“Exactly!  Totally brilliant.  And you want a President who’s that smart … who’s that much smarter than you.  You want a guy who can understand all the stuff the American people don’t.  Because that’s how you fix it.  You may not understand it, but I do, which is why you elected me.”

At this point I coughed.  “There’s a problem with that.”

They both looked at me.  “What?” said the first guy (software development).  “You’ve got a problem with intelligence?”

“I love smart,” I said.  “But this is a democracy, and saying to the voters ‘you don’t understand what I’m doing, but you’re going to let me do it because I’m smarter than you,’ isn’t a legitimate argument.”

“Why not?” said online marketing.

“Because you can’t have an informed electorate if you’re refusing to talk to them in a language they understand.  Maybe you’re that much smarter than them, and maybe you’re not, but if a democracy is going to function, the people have to know what their elected officials are doing.  The smartest man in the world will be a dictator if, as president, he doesn’t have to explain what he’s doing to people because he’s so damn smart.  He has to explain it to them.  And if he can’t, then he’s not a good president.”

Sunkara is making a related point:  that data doesn’t tell the whole story.  That it is an incomplete approach to what it actually takes to govern a nation.

This is counter-intuitive to people who live and breathe data. After all, isn’t data information?  If something is relevant, can’t it be quantified and added to our data pool?

In fact, no.  We cannot quantify everything relevant.  Maybe this is because we’re just not good enough at it yet, or (as I believe) because not everything is quantifiable – but the way in which data is framed, and connected, and interpreted, matters as much or more than the raw information.

Or, as this article cheerleading the application of “big data” to literary criticism sheepishly admits at the bottom:  “A content analysis of Animal Farm can tell you what Animal Farm says about animals, but it can’t tell you what it says about Stalinism.”

I don’t share Sunkara’s disdain for data-driven centrist pundits, who I think serve a valuable purpose, but I do know that the extent to which a political party believes that better data requires the abdication of moral authority is the extent to which they trade the rock of ages for quicksand.