Pinterest's user experience has drawn most of the attention, but the data users generate are what's really interesting.
Ben Silbermann (left) being interviewed at SXSW (flickr/pantavila).
Ben Silbermann is quiet, reserved even. When I arrived at GigaOm's Roadmap conference this week, he wasn't in the speakers room BSing with the journalists and entrepeneurs there. Instead, he was sitting quietly backstage watching Om Malik interview Evan Williams under the bright lights on a small monitor. When I asked him how he was doing, he told me about life with his infant. We both watched a clock count down to the moment when we had to go on.
I mention all this because Ben Silbermann doesn't do a ton of public appearances, or even interviews with journalists. Which means when you've got the guy there and willing to answer questions, it's exciting. In the spring, there were dozens of stories about Pinterest. That's dialed back in the past few months (aside from Fast Company's excellent feature
), but Pinterest just keeps growing and growing.
By now, most people are familiar with the company's mechanic. You can decompose any web page into its constituent images and pin them to one of your "boards." That's the user side of the experience and it's very, very slick. Silbermann contends that Pinterest's core value is that it lets users plan their futures, unlike Facebook (organizing your past) or Twitter (narrating your present). That's how he sees his product fitting into people's lives, he told me.
I opened the interview with perhaps too much of a focus on the demographics of Pinterest. You have almost certainly heard that Pinterest has more female users than male ones. But it's also more Midwestern than your average young web product. I'm not interested in these facts per se, but I would like to know how and why the network developed. Was there something to the core mechanic that disproportionately appealed to women? Or did they just happen to populate their beta network with a lot of Midwestern women and from that seed sprung this whole interesting tree? Silbermann told me he thought it was a little of both.
The question I was saving up, though, didn't have anything to do with the user experience of Pinterest. All the time I've spent reporting on how companies like Google and Nokia build maps had convinced me that building tools that allow you to structure vast amounts of human knowledge into a machine-readable format is an amazing way to create value. This is what librarians do. And this is what Google Translate does. And it's what the people who make the map software on your phone did. The machines are amazing at using the data, but we're the ones who are good at parsing the logic of the human world.
One of the big tasks in artificial intelligence, for example, is labeling photographs. Both Microsoft and Google have built cutting-edge (and huge!) neural nets that can identify cats in YouTube videos, for example
. They are getting better all the time and there have been several step changes in how good they are over the last five years.
But could the big machines separate cats into cute cats
and silly cats
, or recognize a picture of cross-species animal friends
? Not really. And this is something that humans can do effortlessly. We impose categories on things because that is how humans work. And another name for a Pinterest board is a category.
So, if you take this perspective, Pinterest becomes something wholly different. It's a fun game to get users to embed their knowledge about the objects and logic of the human world into a database of photographs.
That's what I really wanted to ask Silbermann about. What's he gonna do with all that beautifully, humanly organized data?
And right as I was winding up to that question, working our way towards it, a fire alarm rang. At first, I told people to hang out in their seats for a few seconds, hoping that it'd switch off immediately. But the clanging went on. And soon, Silbermann and I were making our way down the back stairs and out into the unusually warm night. He had a meeting back down in Palo Alto. I hopped in an Uber car and rode back to BART pinning the images from my day to mental boards: missed opportunities, humans vs. machines, San Francisco summer in November, fire alarms.