The past twelve months have been search-intensive. I mean I have seen a lot of "next generation search" deals. And I know I am far from alone. Socially-enhanced search, tag-based search, machine intelligence search, multimedia search, etc. And I have done a couple of search deals, too.
Nearly every search startup I see is attacking the Googleplex head on with a "better experience" behind the search box. This is flying directly into the buzzsaw, at least for now. Google had a market opening because the first generation guys (including us at Infoseek) took the eye off the ball and became "portals". Maybe this will happen at Google, but I'm not betting on it.
The attraction to search is magnetic. Search monetizes because of the intention behind the act of searching. Search is a great business because Intention is Attention.
So how do I think of search as an investment opportunity in light of the dominance of G, Y, and even M? More and more I am convinced that search is going to become a service, not an experience. For the moment I am heavily influenced by Jeff Hawkins' book On Intelligence. A Great Read. (Thank you, Naval.)
Hawkins' main thesis that that thinking is pattern matching. Intelligence is better, faster, more fine-grained pattern matching. I am not going to try to restate the whole point. Read the book.
But search is pattern matching, too. Search == pattern matching == thinking. Thinking is the process embedded between behaviors. Search can and probably will be, too. Search is not the end user experience any more than typing ">c: dir" or ">ls" was. Search is a service, not an application. By analogy, consider the SQL query in a software application. It is an embedded search process in service of a larger experience. It is an intermediate step to match a well-formed need (request) to a well-formed fulfillment (response). The application logic and user interface are major value-added steps on top of the basic request/reponse mechanism. Of course request/response infrastructure is a big business. Oracle and Google can both attest, being the current big dogs in structured and unstructured search respectively with radically different business models and business practices.
I think "Search Inside" is one direction where search is going. Web applications that embed search to produce a richer experience are going to be really interesting. A better retrieval or ranking algorithm feels like a marginal change not justifying a change in consumer behavior. Machine intelligence with embedded search is going to feel like real intelligence. Are they going to be built on a G, Y, or M search platform or have their own. Don't know.
This is pretty abstract stuff. If you don't really know what I mean, you're not alone. Neither do I, yet. But I'll know it when I see it. And this post is an unabashed troll is find entrepreneurs/technologists who have interesting ideas that relate to this thought.
My only example to motivate the point is this, and the example is pretty trite. There are lots of travel search sites. But I have yet to see a real web travel agent, one that embeds search to fulfill my real goals when I travel. My real vacation goals are about relaxation, interesting things to do, and budget and time constraints. The end result is the constellation of transactions and prepatory content that promise to create that experience. That's what a real travel agent used to do before Travelocity and Expedia disintermediated their commissions. Travel agents embedded search to create a higher value result for the client.
This is the kind of "software agent" example that has been around since the invention of object-oriented programming. It won't happen any time soon, if ever, but interesting enablers will be built along the way.
As I said, we made two investments so far in search. Riya and Vast [no public site, yet]. In retrospect, both are enablers of forms of embedded search. Search can't be a foundation service until it reliably returns results that are both precise and well-formed to enable some value-added transformations/processing. Riya is about automatically associating a data type -- images -- with metadata. The association of metadata (identities, time/place, objects, text, permissions, etc.) makes the image retrieval both precise and reusable in upstream applications. Vast is building a WWW crawling, parsing, and extraction technology for mining the deep web for unstructured content. Its technology is a 'content factory', transforming unstructured content into a structured reusable form -- crudely put, it turns the Google world into the Oracle world. Both companies are enablers. They enable different forms content reuse for the Search Inside vision. Along the way, they will look like interesting end user applications, but they really are being built with a vision of being components of a Search Inside vision for web applications.
I know this begs a lot of questions. And the holy grails (some might say delusions) of AI and the semantic web are lurking behind the curtain. I don't pretend to have the answers. I can only say I believe this notion of search-qua-thinking is an important fragment of an idea. I don't want to invest in better search services. I want to find companies that are going to enable or use Search Inside(TM) (With apologies to Intel Marketing).
PS.
Soon after my initial post, I received several thoughtful replies about the failures of AI, mostly from AI experts. I guess this reads like a call for AI -- not meant to. That's why I consider this a fragment. It is admittedly some fuzzy thinking about fuzzy thinking. But is also way to reach out to people with innovative ideas at the edge, because the center is so dense with replicas of what's successful today.
