Half a century ago, management guru Peter Drucker introduced the concept of the knowledge worker. Today his grandson, Nova Spivack, is trying to turn their knowledge into something more than the sum of the parts and boost their collective intelligence. Spivack is founder and CEO of Radar Networks, which recently launched Twine, which bills itself as the first consumer Semantic Web application. Twine has raised more than $20 million from some big names in the venture capital world, including Velocity Interactive Group, Vulcan and DFJ. The site recently opened to the public and claims 700,000 unique visitors and around 50,000 register users. Here Spivack talks about the shortcomings of Google-era search, the promise of the Semantic Web and his vision for how knowledge workers can become collectively smarter.
—In a nutshell, what is Twine?
Twine is a service that helps you track and discover content and products that relate to your unique interests. It learns about your interests as you use it. It recommends things to you and also helps you share and discover things with other people who share your interests.
—How is Twine different from conventional tagging or information sharing sites?
With other sites, you have to type the tags yourself and do the work of finding things and organizing them. In Twine, the system is smart so it actually learns. As you add information, Twine reads it, understands language and it figures out what the content is about. If you bookmark some page, Twine reads the page and generates semantic tags that have meaning. It knows that John Hancock is a person and in a different sentence it might refer to company. It also goes out and crawl the web for information for whatever you add.
Twine organizes information for you automatically—anything you bookmark, files you send in, emails you send in, even things that you write directly in Twine. Anything you add to Twine gets analyzed, data mined, crawled, and becomes part of your knowledge base. As it learns about you it starts to recommend other related information that you would like. It really is a next generation way to keep track of interests.
— What’s wrong with search as we know it?
A keyword search finds haystacks, but what you really want are the needles. How often do we do a search where it finds 593,000 results—that’s a giant haystack. The needle you want isn’t necessarily there on the first page of results, but statistics show most people don’t go beyond one or two pages of search results. If it’s not on the first two pages, we’re probably not going to find it. Instead, people do a series of queries trying to get the results they want onto the first few pages. Basically, people are hacking Google to try to get a query that gets the results they want out of the first two pages. Google doesn’t really understand what you’re asking for. It’s just trying to match statistically some keywords to some pages.
The other problem with search engines like Google is they tend to favor popular pages rather than the page that has the thing you really want, which might not be such a popular page. In Google, if the page isn’t highly linked from other pages, it probably won’t score very highly and probably won’t end up on the first two pages.
So the first phase of search is "give me the 500,000 pages"—that’s Google. But the second phase is the Semantic Web—let’s actually analyze those 500,000 pages and find the specific needles that match what you really need. That requires intelligence and reasoning.
—Can you briefly explain the concept of the Semantic Web?
The Semantic Web is a set of technologies that enable the web to be understandable by software. Today software can’t understand the web. It just sees data; it doesn’t know what the data means. The Semantic Web provides almost like a markup language for meaning. If the meaning is there in metadata, the software doesn’t have to be so smart.
Being able to specify the meaning of the content of the web actually transforms the web from a file server, just a bunch of documents—which is what the web is today—to a database that all applications can share. The web will become far more searchable, more precise, and more meaningful. It also means software applications can get smarter. They can do new things with the content of the web because they understand what it is better.
—How is Twine a semantic site?
It’s semantic because it actually understands language. That’s number one, linguistic understanding. It actually can read and understand the meaning of any web page, file or email message that you put into it. It learns about your interests. Number two, it actually uses the standards of the Semantic Web, which is an important new stage of the web that comes out of the World Wide Web Consortium. These standards basically enable the meaning of information on the web to be specified so that software can understand it. It makes the web readable to machines. Twine is really the first major consumer application that’s built on the Semantic Web technology stack.
—If we think of the Semantic Web as web 3.0, when do you think it will arrive in the mainstream?
I think these things generally happen in decades. Web 1.0 was roughly 1990 to 2000 and that was mainly focused on the backend of the web. Web 2.0 was 2000 to 2010 and the focus in this decade has been on the front end of the web, the user experience and making the web feel more like desktop software. The next decade, 2010 to 2020, the pendulum swings back to the back end. This is where the Semantic Web will be extremely important, as well as a whole lot of other technology standards.
The basic idea is a new generation of the web, which is focused on upgrading the way the data applications work. The web is transforming from a bunch of separate collections of data and separate applications to something that will be more like one big database and one big operating system that everything is connected to.
Basically the Semantic Web will be a giant database underneath the web OS. Applications will be able to move around freely in the databases. Effectively what we’re creating is a gigantic computer, what Kevin Kelly, the editor of Wired, calls "the One Machine"—a gigantic computer with an operating system, a data management layer, and applications on the top.
I call it the global brain: we’re creating this one giant system which connects all of human data and all human minds, all our machines and all our software. Everything is connecting to this one giant web.
—Why call it Twine?
Twine is like the string. Twine as a service ties together all your interests, all your information, all the different paces you do things on the web, all your content, the people you know and new people you should know who share your interests. You can think of Twine like this connective tissue of the web.
—You’re Peter Drucker’s grandson. How has he influenced you?
My grandfather invented the term knowledge worker. A lot of his work focused on knowledge work, the nature of knowledge work, knowledge organization and the knowledge economy—all of these kinds of concepts he really originated.
We used to talk about this a lot and that was a big influence on my thinking. It got me to think about organizations, collective intelligence and organizational intelligence. That actually is what inspired Radar Networks, our company, and Twine, our product. Basically, Twine is creating what I call "connective intelligence"—intelligence that comes about by connecting people together to be more intelligent. By connecting people in a smarter way, you can facilitate the intelligence that’s already there and amplify it.
I’m interested in facilitating groups to be smarter collectively. How do we enable organizations to evolve to a new level of collective intelligence? This is where I intersect with some of my grandfather’s thinking. The difference is he was really on theoretical side and I’m almost completely on applied side—building things on a very roll up your sleeves level.
—So how do organizations get collectively smarter?
There’s three levels of collective intelligence. The first level is a crowd. It’s like a school of fish or herd of cattle. If you look at that from a distance, it seems like the school of fish is moving intelligently, but there’s no real collective intelligence. It’s just local intelligence acting in mass.
The next level up is a group. In a group, you actually have some form of structure and command and control. You have a sort of nervous system, if you will. There’s some leader or group of leaders. Groups have the beginning of structure.
Above the group is level three, the meta individual. They key to that is when a bunch a parts become the new self, a new entity or individual rather than just a mass. To reach that level of collective intelligence, you have to evolve something that’s like a collective analog to a self. Most organizations if they have that at all, it’s very vague. But I think the next phase of evolution for organizations is to make that concrete using systems like Twine and other new tools to provide a service that acts as this meta self and facilitates this new level of connective intelligence. Basically it gives the organization a next generation nervous system.