Start-Up StumbleUpon Overhauls

An Internet search engine dedicated to serendipity and surprise faces growing pains.

Most mornings, I start online, drinking coffee, checking to see if Josh Marshall of Talking Points Memo has had any stunning 1 a.m. insights. As part of the wake-up ritual, I take a look at my StumbleUpon (SU) page to see who has paid a visit. A number in red at the top of the screen tells me how many people have stopped by, and when I click on it, I see postage-stamp-size images of avatars labeled with screen names.

I scan the column of visitors for people I know. There’s Blackbird, who lives in Zagreb, Croatia; he and I are old friends, although I have never met him and don’t communicate with him much. Our “friendship,” like many on SU, is predicated on a vague overlap between his interests and mine. Hackenbacker, a software engineer living somewhere outside of The Hague, is another old SU friend; he used to e-mail me photos of the strange minimalist furniture he designed in his spare time. We’ve also never met.

This is the nature of StumbleUpon, an eccentric piece of software. A toolbar originally designed as a plug-in for the Firefox Web browser—with a clunky name that betrays its Canadian origins—SU is a search engine designed for imprecision.

I down­loaded it in 2005 because it promised something rare. Unlike other search engines offering specificity, SU simulates what happens in the real world. You walk down a street with a particular des­tination in mind, but along the way you stumble on something unexpected. Most Web tools are intended to give you what you think you want, or what it determines you want; StumbleUpon is set up to provide happy accidents.

The basic interface is simple. You hit a button labeled “Stumble!” and the software sends a Web page that corresponds to your stated interests. Then you click on a “thumbs-up” or “thumbs-down” button to rate it. Your actions (approve or disapprove) influence the likelihood that others will see the same page, and affect your own search results the next time you hit the “Stumble!” button.

Of course, it’s not that simple, because Stumblers also get a blog on which they can post Web sites they like, rant about the state of the world, or quote their favorite poets. These pages attract visitors who can send messages or become your “friends.” So it starts simple, but over time it grows strangely dense, rich, and complex.

I had never written about SU or even examined it with a professional eye until recently, when confronted by the specter of a major interface redesign, I spent an hour talking with its cofounder and chief architect, Garrett Camp. A 30-year-old software engineer educated at the University of Cal­gary, Camp began creating an Internet start-up in late 2001 with a group of friends. “To start, it was one button to Stumble and a ‘good’ button and ‘bad’ button,” he told me. “We took all the sites we knew and started placing them into categories. At the time, we probably had twenty categories and 500 sites. The first version went out to friends in February 2002, and it was super-small.

“Then we figured, if people were giving thumbs-up to sites, it would be cool to build a review site,” Camp continued. “So we added a review page where you could see people giving a thumbs-up. Then came user names, then pictures and profiles, then messaging, and after that we said, ‘Oh, there should be friends.’” In theory, networking plays a secondary role to discovering cool Web sites. “We added the social features because we thought it made it easier to get interesting content.”

Three years ago, with their site hosting 600,000 users, the developers relocated from Calgary to San Francisco, raised a couple of million dollars, formed a real company, got serious about advertising, and began to grow. (Ads are Web pages that someone has paid to inject into the Stumble stream.) In 2007 SU was acquired by eBay. There were dire predictions. Some Stumblers feared that the weird ecology of SU would be contaminated by its association with the online-auction site. A few people quit, but in truth, nothing much changed.

Still, I wasn’t too surprised when I began receiving urgent messages late last year about an up-coming overhaul of StumbleUpon. I kept receiving copies of a screed by a Stumbler named Take­me­­awayplease, who declared, “I strongly believe that SU should now be classified as SPYWARE.” He alleged that the upcoming redesign would turn the toolbar into a mechanism that would log all Web activity instead of just pages that one rated with thumbs-up and thumbs-down. “I think it’s just paranoia,” Camp said. “Our toolbar is completely open sourced. We have a ton of engineers at Mozilla who’ve looked at it and trust it.”

More disturbing was a post by one of my SU friends, Saline, a journalist based in Australia with a fondness for bird and dog photos. She characterized the redesign as “a horror story of embedded marketing. It is…the nightmare exploitation of the so-called ‘men in suits.’” I asked her to explain, and all she could say was that the “new owners”—eBay—have a “predatory” agenda.

At that point, I contemplated sending a farewell message to my favorite Stumblers and uninstalling the toolbar. But when I started examining the new design, it seemed crisper. There was more white space and fewer of the goofy little icons that gave SU a homemade appearance. A newly added set of buttons led to collections of Web pages on specific topics such as “arts” or “health/fitness.” I suspect that these categories might become venues for advertising, but this hardly rises to the level of “a horror story of embedded marketing.”

I asked a couple of my StumbleUpon designer friends for their take on the overhaul. Most of them found bugs and buttons that don’t quite do what they’re supposed to do, and some of them were annoyed by the aesthetics. Hackenbacker believed that the cleanup made his blog look “cluttered.” But Donna Mugavero, a Bethlehem, Pennsylvania–based graphic designer who Stumbles as Brickgrrl, pinpointed the real source of the anxiety. She recalled working at noncommercial radio stations where the DJs would shun cherished alt-rock bands like R.E.M. when they achieved mainstream success. “I think SU is like that,” she said. “New for­mats mean more public accessibility and the resul­t­­ing wail of ‘It’s nothing like the old days!’”

Perhaps the disgruntled Stumblers sense the truth: the redesign is largely about growth. Camp said the “whole reason” for the overhaul was speed, efficiency, and increased security. The most sig­nificant new element is a Web-based toolbar that will allow anyone to Stumble, receive random pages, and rate them, without having to download a thing. “We’ve realized, if we want to get to Facebook or MySpace size, we’re going to have to make it eas­ier to do. To get that next hundred million people, we need to make it more accessible,” Camp explained.

“Why a hundred million?” I asked. “I don’t know,” Camp said. “I just think it’s a good round number.” He argued that as the Internet takes over from television as the dominant medium, the ability to find content you didn’t know you were looking for becomes crucial. Perhaps, but the rates SU charges advertisers “start at five cents per targeted visitor.” Multiply five cents by 100 million and you get a very round number, indeed.

The most surprising piece of news that came out of my talk with Camp was not that he wanted to grow SU to 100 million but that it was already pushing seven million. StumbleUpon is fast approaching the size of New York City. And maybe it works the same way. When you live in New York, you reap the benefits of the multitudes—cultural richness, limitless opportunities, exotic food—but you don’t feel the bigness because you mainly interact with people you know. I was initially horrified by the notion of StumbleUpon’s being as big and well-known as Facebook, and I’m tempted to use the clean new design as evidence that growth engenders blandness. But when I think about it, I have to admit that it’s difficult to see a downside to having millions more people feed an engine that generates serendipity.

Recent Viewpoints