Twine, a bookmarking and knowledge-tracking application developed by Radar Networks, is leaving its closed beta tonight with an array of new features, in hopes of becoming one of the definitive websites of the next generation of the Internet.
Here’s why it might succeed: It hopes to create a web of connected knowledge encompassing the whole Internet, which will start by replacing your bookmarks bar and finish by assembling a modern-day Dewey Decimal System for the cluttered, often unnavigable mass of information on millions of web pages. (Of course, it might share that responsibility with other applications, or fail entirely — more on that in a moment.)
The way Twine works bears a strong resemblance to popular sites like Del.icio.us and StumbleUpon. Users submit a link and a short summary to specific interest groups, which are usually more or less categories; this article, for instance, might go into “Startups”. Other members of the same community, or Twine, get alerts or have access to those links. Twines can either be open to the public or private, with only a few users, or even just one.
So far so good. But Twine is built on years of work in semantic technology, a branch of computer science that makes human-generated information understandable to computers. So with every link or other bit of information submitted, Twine goes through and “reads” the text to pull out meaningful tags. So for this article, it would know that I wrote it about Twine, and that Twine can be referred to as either a bookmarking or interest tracking application.
Each article tends to generate a rather large set of tags, for each of which Twine also attempts to guess their importance and relation to other tags. What builds up, over time, is a densely cross-referenced knowledge base, so when you go looking for a bit of information, say “bookmarking”, you’ll pull up all the articles that a particular community ever found important, ranked by their relevance.
I’ve been playing with Twine on and off since it opened to journalists earlier this year, and it definitely has potential. But while the tagging already seems more advanced than even more cutting-edge apps, the interface was a bit clunky and slow. It’s now getting a revamped interface, a significant speed-up, better internal search and a number of other tweaks. It’s also opening all public Twines to search engines and casual viewers.
At the moment, search technology is the only real way to navigate around all the information on the Internet — in-browser bookmarking becomes too confusing after more than a few dozen bookmarks. Meanwhile, social bookmarking sites like StumbleUpon, although good at serendipitous discovery, don’t do a great job with finding specifically what you want. So there’s definitely room for something like Twine to make it big.
As to what could go wrong, of course, there are multiple potential trouble points. Semantic technology is new, and Twine is working on a specific platform, so it remains to be seen how it will scale up. So far, it has only had at most 50,000 active users, according to founder Nova Spivack.
It’s also possible that semantic technology won’t be better than human editors, which is the hope of companies like Mahalo and Search Wikia, who employ people to build information pages and search results. On the other hand, Twine’s scheme for organizing information could involve too much human interaction, in which case a more stable database like Metaweb’s Freebase, a sort of semi-automated Wikipedia, might prevail. Other, as-yet-unknown companies could appear and do a better job at the same thing.
But the most likely problem would be that it’s just too early for something like Twine. About half of web users don’t even use search engines yet, so it might be a symptom of the Silicon Valley bubble to assume large numbers of people want an information tool like Twine.
But just in case you’re interested, go check Twine out here.
VB’s research team is studying mobile user acquisition:
Chime in here, and we’ll share the results