Photo: Flickr/David Gohering
For about a year now, Twitter has behaved very aggressively toward certain kinds of companies in its ecosystem – mostly client-makers, ad networks, and URL shorteners.It all started back in April 2010 when Twitter’s top shareholder and then most-influential board member, Fred Wilson, wrote a post telling entreprenuers in the Twitter ecosystem to stop “filling holes” with products and features that Twitter was likely to someday build itself.
Today we learned what kind of problems Twitter wants other companies its ecosystem to actually work on, when, this morning, Twitter announced an alliance with New York company called Sulia.
Sulia organizes tweets and tweeters into topic streams. It shares these streams with publishers. Sulia is Twitter-sanctioned, and its working with an impressive list of publishers, including The Washington Post, WSJ.com, TweetDeck, BeliefNet, and Silicon Valley’s favourite content-aggregation app for the iPad, Flipboard.
Cleanly organising Twitter by topic is an important step toward finally monetizing the ecosystem. It’s a lot easier to sell a BMW ad against tweets about automotive topics, when you know those tweets are going to be useful and relevant (and not spammy or offensive).
Sulia charges publishers some kind of fee and pays Twitter. Maybe this tax will someday turn into real money.
In a canned quote from Sulia’s press release, Flipboard CEO Mike McCue crowed: “By introducing Flipboard readers to important independent sources curated by Sulia, we help people discover great content, photos, and videos on topics they care about most.”
So how does Sulia work?
Sulia CEO Jonathan Glick explained over email:
There are about 80 million Twitter Lists. These are the only reliable place where User X says User Y is a relevant source on Topic Z. The trick is figuring out what a list is about, ie. what topic Z is, because obviously people don’t label lists neatly with names like “venture capital.” They call them “investors” or “vcs” or “funding bigshots” or “sandhill road.”
So the way we figure out what a list is about, what the topic is, is by doing a realtime analysis of the content expressed by the tweeters in the list, and extracting the most common unusual words and phrases. We can then compare them to the common unusual words and phrases used within our categories and make a match.
As in: a-ha, these folks are talking about “seed rounds” and “super angels” and “tranches” and so on, this List is a venture capital list and therefore the people listed considered relevant to VC.
Then we look at all of the VC lists, and find the most-listed person, who are (drum roll): Fred Wilson, Mark Suster, Josh K, and so on.
What’s REALLY cool is that we can then use those unique words and phrases as the basis of a filter. So we can know to EXCLUDE tweets that don’t include those VC-related words, because by definition words and phrases used uniquely by a group of subject-matter experts are on-topic.