Google a common query, such as “movie tickets,” and you’ll get 55 million hits. The first few — Fandango, AMC Theatres — are legitimate. Then they trail off, until by page 15, you’re awash in come-ons like “Get Your Hannah Montana Concert 3D Movie Tickets!!” (Alas, I’m too late.)
Such chaos hits home for Nova Spivack. Before he started his own company, Radar Networks, the 39-year-old entrepreneur and grandson of late management guru Peter F. Drucker, had so many virtual folders and bookmarks, he’d routinely lose track of important messages and links. Since then, he has watched social news and bookmarking services such as Digg and Delicious (the latter sold to Yahoo for a reported $30 million) garner avid followings for helping people find and store new information.
Now he’s trying to leapfrog them with Twine, his company’s first product. The next-gen bookmarking service has already attracted $26 million in venture capital and is rapidly adding users for what Spivack deadpans is “Digg … on Ritalin.” Twine’s unique visitors have grown more than 40% each month since its October 2008 debut, topping 80% in February 2009 — more than 1 million uniques. It’s also comfortably outpacing more buzzed-about startups such as FriendFeed and mirroring the early development of a little venture called Wikipedia.
Twine’s growth is a testament to the need for a news-tracking service that learns the more you use it. At first glance, the site resembles a souped-up RSS feed. Threads, or “twines,” are centered around specific ideas (“social media”), people (“Barack Obama”), and events (“SXSW 2009”), and users fill them with content found around the Web. The site then tracks the articles they add and the topics they follow, and assembles an interest-based personality profile. Based on what it learns, it sends news and friend recommendations users didn’t even know they wanted. For example, because I was following twines about microblogs and pop culture, I got a story about Britney Spears surpassing President Obama as Twitter’s most-followed person.
Creepy? Perhaps, but it’s an early sign of what the geek squad calls Web 3.0, a “semantic Web” where sites can understand the quirks and relationships in the data they mine, much in the way that humans differentiate between cheesy nachos and cheesy pickup lines.
Web 3.0 and Twine are in their infancy. Today, the Internet caters to humans, not machines; accordingly, Web expert Clay Shirky says, “it’s a mess.” To fix it would require all data, from flight times to photos, to be stored in a standard way, so search engines could understand what they crawl.
Spivack’s team is trying to balance the Internet’s human side with artificial intelligence. Using “ontologies,” or programming algorithms, they load user submissions with metatags to make the data understandable to both search engines and Twine itself. Because the 32-person team can’t possibly program — let alone comprehend — everything about everything, its coding skims the surface.
By the end of the year, though, Spivack plans to make it easy for users to create algorithms and “teach” Twine even more about specific topics, from Starbucks to Star Wars. He’s also tinkering with revenue models that would let advertisers target specific interest groups. As Yahoo and Google start embracing semantic coding, Twine’s efforts could streamline Web search as a whole, all while driving traffic. Already, roughly 30% of Twine’s visitors arrive via search.
At this pace, the site should eclipse Delicious within months as the Web’s premier bookmarking service. Of course, Spivack has probably known that for weeks. He’s following “startups to watch” on Twine.