The Verge and The Huffington Post attempt the impossible: making comments smarter
At the start of December, a reader of The Verge from Stamford, Connecticut, did something kind of dorky but also kind of awesome. Inspired by Nilay Patel’s video interview with Senator Al Franken, the reader, who goes by the username zahntron, took himself and a camera to the office of Congressman Jim Himes, a Democrat of Connecticut’s 4th District. Zahntron then interviewed the congressman about his views on tech issues, such as the Stop Online Piracy Act. “Nilay's interview with Al Franken really inspired me to reach out to my Congressman and find out where he stands on these issues,” zahntron would write on The Verge’s “Meta” forum, where he posted the resulting video. (Oh hey – looks like we just took turned the meta up to 11. Go wild, postmodernists.)
The Verge’s editors would pluck zahntron’s forum post and promote it to the front page. It performed about as well a story produced by Verge staffers would do. In fact, going by the public metrics, it did better than Patel’s original piece. While the Franken interview elicited three comments, 10 Tweets, and 20 Facebook Likes, zahntron’s effort won him 37 comments, 66 Tweets, and 15 Facebook Likes.
The Verge’s forums – or, if you’re kind of a dick, fora – are a study in how a new take on an old form can elevate user-generated content (UGC) to another level. After all, nothing on the Internet is more old-school than the bulletin board. Somehow The Verge has made bulletin boards cool, and relevant.
One of the promises of the Internet has always been that it would blow apart the conventions of the one-to-many broadcast model and facilitate a many-to-many dialogue. That promise has been only partly fulfilled. While the likes of YouTube, Reddit, and Quora have demonstrated that it’s possible to build media products, platforms, and businesses around UGC, in other instances the results have been far less than satisfying. For the most part, that has been true of comments sections and discussion boards, which on balance do add value to media websites but are often dragged down by excessive noise, inania, profanity, insults, trolls, spam, anonymous usernames like [email protected] (40 points if you get the reference), and jackasses who write comments like: “You do know that xxx is xxx, right?”.
The Verge has a comments section, of course, and it is a lively one. But it is the forum that it holds up as a higher ground for community conversation. It has designed each forum comment page so it looks like a story, replete with social sharing buttons, its own comments thread, a section header, and boldface headline typography. “We decided we wanted to do forums when we launched the site because we wanted to be more than just a news site,” says Dieter Bohn a senior editor at The Verge. “We wanted to build a real community.”
Bohn believes that contributors are more incentivized to produce good content because of the pretty format. That doesn’t mean all content is better than a standard forum, Bohn says, “but we get higher peaks than you would get on a standard forum.”
Forum contributors, like zahntron, sometimes get star treatment. The Verge editors actively scour the boards for content worthy of promotion to the front page, which has so far included an explainer on touchscreens, a demonstration of what a 4-inch iPhone would look like (four months before the 4-inch iPhone 5 was announced), and an essay about the iPhone 5’s app switcher.
On a broader level, The Verge’s design and editorial decisions around forums can be understood as an attempt to filter signal from the noise of UGC. One of the main problems of user-contributed content on big media sites is often not even that it’s low quality, but that it’s too abundant. The Huffington Post is another such sufferer of the comment-overload affliction. Take a look at its lead story right now: already it has way over 2,000 comments. Who’s supposed to read all those? There may be a few worthwhile comments in there, but how the hell do you find them?
Well, the HuffPo has been thinking about those questions too. It recently launched a new platform, Conversations, which uses machine learning to find and highlight the best contributions from readers within comments sections. Each comment is attached to its own URL and can be shared to Facebook or Twitter. The platform – which is so far live only on the World, Books, and Gay Voices sections – comes courtesy of the Huffington Post’s 2010 acquisition of Adaptive Semantics, a two-person startup that specialized in semantic analysis. The HuffPo’s tech team has spent two years since the acquisition building on Adaptive Semantics’ work. Conversations is based on an analysis engine called JuLiA, which searches comments and looks for linguistic and structural clues that signal good feedback. It’s the same technology the company uses to auto-moderate comments in real-time for its video streaming site HuffPost Live, and it comes in handy for dealing with the 100,000 or so comments that the publication gets each day. In its lifetime, Huffington Post has hosted more than 225 million comments. (For more on Conversations, see Jeff John Roberts' earlier piece over at PaidContent.)
“We want to bring to the fore the most interesting discussions that people are having,” says CTO John Pavley, who describes the HuffPo community as “first-class citizens” within the organization. “It’s almost like every article is a town hall.”
Indeed, community discussion appears to be undergoing a mini-renaissance on media sites in general. Nick Denton has made Gawker Media’s new discussion platform the focus of its sites, encouraging advertisers to take part of the conversation surrounding each article and placing almost as much emphasis on the comments as the original story. He has said many times that he thinks of comments as content, not data attached to content. And TechCrunch recently switched from Facebook Comments to Livefyre, a more flexible and dynamic system that allows anonymity, in an attempt to stir up more discussion around its stories. TechCrunch’s move, however, actually goes against the trend of “less noise, more signal” that The Verge, HuffPo, and Gawker are riding.
For those three properties, the thrust towards improving discussion systems is an attempt to not just extend but also amplify an article and its import. So far, comments – despite, and because of, their abundance – have failed to live up to their potential on that front. But thanks to machine learning, semantic analysis, design tricks, and a heavier curatorial hand, 2013 might just be the year when user-generated content on media sites finally gets smart.
Upvote if you agree.
[Image courtesy Ray Fenwick]