Tesla in a Teapot
I've been following — as who hasn't been? — the fight between the New York Times and Tesla Motors. A recap:
First, New York Times journalist John Broder reviewed the Tesla Model S electric car and had a problem with it — namely that it couldn’t hold a charge.
Second, Elon Musk, founder of Tesla Motors, retrieved the log files from the car — apparently everything the car does is tracked and stored as a set of data points — and made some charts and graphs. He disagreed with the Times, accusing that august outfit of base deceit. He titled his blog post “A Most Peculiar Test Drive.” I wish he hadn’t called it that, because it made my lips purse so hard I half-fainted. But Musk will do as he will.
Some disclosures: I write little things for the Times (more often memos for internal consumption than things for publication), and I hate cars. But I don’t mind electric cars or biodiesel.
Third, after Musk published his post, netsharks smelled blood in the ether and threw the typical collective fit. Tribal alignments were exposed: Defenders of journalism stood up for the Times. Defenders of log analysis, electric cars, and/or Musk took the entire concept of journalism to task. The kerfuffle demonstrated that technologists who discuss journalism are exactly as stupid as journalists who discuss technology. Much of this transpired on message boards and Twitter. (Doubtless there were actual, serious discussions happening in rooms with closed doors and anxious investors; these probably were much cooler and involved more yelling into those starfish-like Polycom conference phones.)
Finally, at the end (one hopes) of the affair, nothing had resolved. TheTimes public editor weighed in with a sensible-sounding assessment, saying, essentially, “Our dude should have taken better notes but come now, this car is busted.” And of course she pointed out that the Timesshould be more accountable than some guy writing in a little notebook. Little notebooks won’t cut it in the age of mass analytics.
Log files, man. Log files.
I work in the storytelling industry, mostly as a hobby. And the thing I’m learning from people outside of that industry is this: if people in the storytelling industry don’t reduce everything to numbers, we don’t stand a chance. “Readers” may buy stories, but everyone else wants numbers. Circulation, subscription, CPMs, RPMs, eCPMs, etc. It all translates, at some level, back to money. That’s the metric underneath all the other metrics.
I used to check my stats religiously. I used to know where people came from and what they did once they got onto my various sites. There was a certain joy in posting something and watching hits roll in. Whoa! Australia! The Arctic Circle! Inside the basement! Reducing things to a number is fine; it is inevitable, in fact, when the primary way that we store and experience information is through computers. Anyway, there is this new cultural logic in which everything must be a number in order to function with all of the other numbers.
In an essay titled “Whom the Gods Would Destroy, They First Give Real-time Analytics,” Dan McKinley of Etsy writes:
Depending on the change that's being made, making any decision based on a single day of data could be ill-conceived. Even if you think you have plenty of data, it's not farfetched to imagine that user behavior has its own rhythms. A conspicuous (if basic) example of this is that Etsy sees 30% more orders on Tuesdays than it does on Sundays.
This explains Nate Silver, Google, OKCupid, and many other things. So of course a Tesla car is a rolling web server. That’s the most sensible thing for it to be. Eventually we’ll be able to merge all of these numbers together. Before long OKCupid will be able to tell you exactly how likely it is that a Tesla owner will hook up at a charging station near Boston. All of a sudden log files, journalism, etc., are in the same cultural bucket. Everything is the Billboard Hot 100.
Editorial people used to take me aside and just yell at me about the web. As if I were somehow the web. “This is terrible!” they’d say. “We can’t allow it!” In my career I’ve had a lot of people yell at me as if I were the thing that was changing. Now they ask me if I know any data journalists. Or PHP programmers.
God knows we need more data journalists. That gun thing, in which theJournal News in upstate New York put all the gun owners on a map, with their names and addresses: it was bad design. I don’t have any trouble with public documents being exposed more widely, but the paper put its imprimatur on messy data. It didn’t need to show the addresses. It just needed to shade in which neighborhoods had the most guns. That would have been more accurate, even if it had been more vague. And it would have left everyone guessing, and a little anxious, about the guns in their neighborhood. Anxiety would have sold more papers or generated more traffic.
I increasingly think that traffic is a side effect of anxiety. One of my business partners is a great banner ad developer. I know you may not believe there is such a thing as greatness in banner ads, but you are wrong. You have seen his work, and there is a 0.1% chance you have clicked on it.
I’ve been watching him for years, and I’ve come to realize that the way he gets clicks is by making the ads ever-so-slightly annoying. Your mouse sort of drifts over them and then you feel this kind of itch. And when you go in to scratch the itch he makes something happen: a balloon inflates, a lunchbox opens. Well maybe I wanted to click after all,you think. I mean, what’s inside that lunchbox after all?
I want to create a website called This Day in Anger. It would just be a list of what everyone is angry about that day. But one of the middle-sized insights of my career in media has been that optimizing for traffic is not the same as optimizing for power.
Another middle-sized insight is that the web can’t resolve the enormous amounts of anxiety it creates. Whereas books and movies create anxiety, and then they drive through to a conclusion.
And so I keep thinking about the website, the content platform, the system that would generate traffic without stress, that would help people resolve something. And then I think: I wish I had the time to write a book about that.
If I were to take you aside and whisper the words “ethical analytics,” you’d just laugh at me. You’d say, “Whatever.” Does anything sound more stupid than “ethical analytics”? Nothing. Nothing sounds more stupid.
Log files or journalism? Yes! Stories or stats? Sure! Tesla or Times? Why not?
We know that stories are unreliable. But so is data. You can’t just throw up a graph and say, “Here’s the truth.” People on the Internet won’t let you.
It’s hard not to see trends, though. It’s hard to avoid interpretation. “We’ll be profitable in weeks”; “The virus is definitely not going to spread any further”; “At this rate I’ll lose a hundred pounds in three months.” It’s the “at this rate” that gets you.
You’re being logged right now, of course. Somewhere a disk has noted down that you’re reading this essay; the image above is the stats view of this story. I am in the storytelling industry, you are in the story-reading cohort, and we meet together in the great temple of analytics. If you made it this far then you know we have that in common. Thank you for reading. And hello!