The Future of Search and the Intelligent Web, From Vulcan Capital’s Steve Hall
Search is hot again. Just when you thought Google had everything figured out, along comes Microsoft with its new search engine, Bing. This upstart effort may not supplant the search king, or even compete with it all that seriously, but it will certainly make things interesting for a while.
Even more interesting, though, is the batch of new technologies and businesses bubbling up around the edges of traditional Web search. Take software that helps you understand relationships between people, places, and products online. Or Web services that try to collect content you are interested in, let you share it with others, and keep you up to date with the latest news about people you know. Google itself is getting into the act with its new Wave tool, which integrates e-mail, instant messages, documents, wikis, tweets, and other social media—all the comforts and stresses of modern life, really—in one place.
All of this raises the broader question of how we’re going to deal with all these different data streams on the Web and use them to improve our work and personal lives, rather than waste time doing an endless (and growing) number of searches for relevant information. To hear more about this emerging space, Luke and I talked recently with one of the field’s leading investors, Vulcan Capital managing director Steve Hall. (See also our wider-ranging interview with Hall and Vulcan Capital president Chris Temple here.)
In recent years, Hall has led investments in companies like ZoomInfo, Radar Networks, Evri, and Gist (the latter two are in Seattle), which all play in an arena he calls “information discovery and the intelligent Web.” Hall emphasized that “Google is not the be-all, end-all for how we discover information.” Indeed, in a deep dive into the technology and business opportunities of this fast-moving space, he talked about how companies can “rewire the Web” to be smarter, how to beat today’s search engines and help people find information more efficiently, and why World Wide Web pioneer Tim Berners-Lee is wrong about the “semantic Web”—his original notion that computers will someday be able to understand all kinds of content and transactions online, if we help them by tagging things like photos and documents with descriptions of their meaning.
An edited account of our discussion follows:
Xconomy: What is your background in online search and the intelligent Web?
Steve Hall: One of the themes I’ve been involved in, really since I started doing venture in 1998, has been information discovery, search, and evolution of Internet data. The first venture deal I ever did in 1998 is now called About.com. It used to be called The Mining Company. At the time, our thesis was that the amount of data on the Web is going up and to the right. It’s hard to find the needles in the haystack. Yahoo, Excite, and Lycos were portals, a main interface to the data, but they were largely a taxonomy directory of websites. You didn’t have very good needle in the haystack searchability. You had AltaVista, which had started, but didn’t have real relevancy control to it—there was no quality output on AltaVista.
Our thesis at the time was if you had humans, experts on many topics, comb through the Web for the best of certain topics—video games, pets, gardens, you name it—it would be a new, almost editorial layer to the Web itself. To curate and provide a great starting point for different vertical categories. That ended up being a very good investment. We invested at a pretty early stage, put in $3 million at the time, and as we recall back then, things moved a lot faster than they do these days at the market level. So it went public in late 1999. Eventually, after it was a profitable public company, it got bought and owned by the New York Times. The point there is that since that deal, all the way up to several deals we are involved in today, [the theme is] information search, information discovery, and how we access what continues to be an up-and-to-the-right growing amount of data.
X: Did you look at investing in Google back in the day?
SH: No. For me, I was only East Coast at the time. It was not something that crossed our path, but obviously it worked out well for folks. When you think about Google, it plays up some of the opportunities we see today. Google was the first company that really combined scale algorithms to index the Web, and had the granularity and needle-in-the-haystack keyword search, with human intelligence, which they were using through the form of editorially embedded hyperlinks to determine the quality and relevance of a website. The rest is history. They had the best of both worlds in quality and scale.
As we talk about deals at Vulcan, we sometimes talk about semantic Web, intelligent Web. There really is this emerging sort of evolution of data to be smarter, intelligent information. The machines and the algorithms can increasingly do more of the heavy lifting that we otherwise have to do. You type something into Google, you have to do the sifting through thousands of websites. Google is a great starting point, and we don’t think that’s going to change anytime soon. But if you can tap more technologies that do more of the work for you—one of the big opportunities to beat search is not search at all, but to reduce the need for search. By driving more automated discovery. Search is active. If you can have passive-based, intelligent discovery of information, you reduce the need to Google.
There are new opportunities there. We’re seeing that for sure with our investment in Evri, with our investment in Gist, and we also funded a company called Radar Networks in San Francisco which has a product called Twine. All of this is about “bring the data to me.” In an information-overload environment of the Web, if you can bring it to me in a way that I don’t have to think about, I can stay informed, I can discover, I can find what I need to know and not have to do all that work.
X: Where does all of this fit with the “semantic Web” framework, which has been around for almost a decade?
SH: Semantic Web or intelligent Web is a core theme that I’ve been excited about, and Paul [Allen] has been excited about. For whatever reason, we collectively, and some other folks internally, see the opportunity similarly. We’ve really been passionate about, “How do you drive innovation?” No matter how you slice it, Google is not the be-all, end-all for how we discover information. We’re not trying to out-Google Google, but there are new opportunities for how you drive these things.
Beginning in 2004, we really had a full landscape of the semantic Web theme. We spent a lot of time whiteboarding and brainstorming and prototyping different areas. How do you jumpstart it? How do you get it started? If you go a little deeper on semantic Web stuff for a moment, everyone talking about it assumes some new standard will emerge, and everyone will encode it into their pages, and the Web will suddenly be a smart corpus. [Computers would be able to understand the meaning of data and documents on the Web---Eds.].
That’s Tim Berners-Lee’s view of the world. That’s never going to happen. The Web is too big. There’s too much to retrofit. No one can agree on a standard. And it’s too much work. It’s hard enough to get people to put hyperlinks in.
X: So tell us about your approach to the problem. (At Xconomy, we’ve reported extensively on Evri, which helps you browse information by making connections between entities on the Web, and Gist, which integrates your e-mail inbox with blogs, articles, and social media to keep you informed about your contacts.)
SH: The birth of Evri was sort of an “aha” moment that we had through the back and forth moments. If you could up-convert the Web itself, by using smart, scalable algorithms that understand, “This is a person, this is a place, this is how these things are connected, this is the meta-data about those things”—you can effectively rewire the Web at the surface level. Such that any information you’d want to know at any moment in time is a click away. In effect, convert the Web into a truly browsable interface, versus a searchable interface. You reduce the need to keep going back to [your inbox or search box] by bringing contacts back at the moment you need them. If algorithms understand that at a foundational level, there are big opportunities around that. That’s the genesis of the thinking. Paul was in every one of those conversations.
X: How does this kind of information discovery relate to what’s going on with social media more broadly?
SH: My current fascination is the catalyst that Twitter has been to convert Web media into almost a stream, or feed-based view, of the world. Granted, all of this today—Twitter and Facebook feed views—are through the lens of other people. Facebook is mostly social stuff, and Twitter is a bit more business. But with Twitter, it’s about this person who I respect just wrote about something, so now I should follow this link. That’s discovery for me. I’m not Googling for it. I’m not going to some start page or a media site. I’m just reading links based on my network and people that matter to me. In some cases, they are re-tweeting other people. So this human filter becomes this highly valuable, highly accurate lens of data—it is sort of our new model for consuming media. Which I’m sure you guys think about, being in that business.
X: So let’s sum up Vulcan’s role in all of this. Where is search and social media headed?
SH: When I think about how it relates to us, as it relates to Gist and Evri, it is very thematic. How do you solve information overload? Can you use algorithms to do a better job of bringing the data to us? In Gist’s case, we’re still early days, but we figured out a bunch of stuff that matters to you because you let us, and now we’re giving you a feed. Just turn it on, dial in, and it’s coming at you. You can consume it or not, but it’s all there for the taking.
In the case of Evri, and I’m not sure Neil Roseman [founder and CEO] has alluded to all of this yet, but what you’re going to see is, we’ve got Evri pages and widgets and so forth, but what we’ve really got is 3-plus million feeds being created on the fly by semantic parsing of the Web as it comes through in real time.
Imagine if suddenly you don’t need to rely on a single person you’re following to get an interesting feed. Imagine you can follow anything you want. Even any category of thing, because we semantically understand that. I might want to follow Seattle journalists, not specifically, but as they are referenced in the media. I might want to follow a quarterback, or a band, or a song, or so forth. These all become portable, subscribable, smart objects that can be passed around and consumed in different applications. So if you think of Tweetdeck for example, it’s symbolic of a new Netscape. It’s a new window into a new form of content that didn’t exist a year ago. What sort of new things will feed that ecosystem? We think of Evri as being a new source of content on these new feed-view platforms we’ll bring at you. It will radically change how we think about consuming media.