0

The Future of Internet Search

NEW YORK – Imagine that Googling an address gave you a list of the closest buildings, ranked by distance. Not exactly what you were looking for, most likely. But that is pretty close to what we still accept for most Internet searches. You don't get what you actually want to finish your task; you get a list of pages that might lead you to it.

That is beginning to change. Even as the online world has turned its attention from searching to social networking, search is getting interesting again.

Consider the development of online search in the broadest terms. First came Yahoo!, with its carefully cultivated (by human editors) catalogue of interesting web pages. Then along came Google, with co-founder Larry Page’s innovative ranking of Web pages not just by their content, but also by the quantity and quality of other pages that link to them.

Social networking brings a new insight. People are likely to buy what their friends recommend, which is why marketers should spend time on social networks and join the conversation, rather than interrupt it with traditional advertising.

But what happens when, influenced by their friends, people actually go to buy something or take some action? That long list of blue links to pages that may or may not contain what they want looks pretty old.

Now, however, something is happening to fix this, and it’s not just a prettier background. It’s structure – the same sort of context the old Yahoo! catalogue supplied, but this time automatically generated and deeper – and across more than just a few categories such as sports and travel.

For example, what people want (and are now getting) in  product search is not a list of pages, but a set of products displayed in some meaningful fashion. They want a map of the product space, not a list. The challenge of course, is that each kind of product has a different structure and a different set of attributes.

Consider wines: you can sort them by price, year, or region of origin, by red, white, or rosé, or by sparkling or still. For clothes, you want sizes and colors – and perhaps some filters depending on your personal characteristics – and of course a “buy now” button.

Some areas, such as travel, are even more complex. To “map” travel properly, the software needs to understand such things as time zones, flight duration, layovers, and the like, along with concepts such as coach or first class, deluxe and standard rooms, double vs. single, and so on. That is why there is a whole separate vertical market for travel, but one that first Bing, and now Google (with the acquisition of ITA Software), may be claiming.

For a long time Google didn’t need to do much to remain the leader in Internet search, focusing primarily on the “access” part of its self-proclaimed mission “to organize the world’s information and make it universally accessible and useful.” But runner-up Microsoft went out and bought Medstory in 2007 and then Powerset in 2008. (I was an investor in both of them, so I have been watching these developments with interest, but I have no inside information on either company since the acquisitions.)

Medstory has a deep understanding of health care, including the relationships between diseases and treatments, drugs and symptoms, and side effects. Powerset, a tool for creating and defining such relationships in any sphere of interest, is broader but less deep.

This all happened a couple of years ago – just before Yahoo! gave up on search entirely and handed that part of its business over to Microsoft. Also around that time, Bill Gates uttered one of the smartest things he has ever said: “The future of search is verbs.” But he said it at a private dinner and it never spread.

To me, the meaning was clear: when people search, they aren't just looking for nouns or information; they are looking for action. They want to book a flight, reserve a table, buy a product, cure a hangover, take a class, fix a leak, resolve an argument, or occasionally find a person, for which Facebook is very handy. They mostly want to find something in order to do something.

As a result, Bing launched a few forays into vertical integration. And in the last few months Google has begun to react. First, it bought ITA Software, which provides the underlying data to several travel-booking sites (Hotwire and Orbitz) and to Kayak, as well as to Bing. Most resellers, a little nervous about Bing’s tool that sends users to book directly with airlines and hotels, are even more concerned about what Google might be up to.

Then, last month, Google acquired Metaweb and its user-generated database Freebase. While Powerset was a tool for understanding natural language and for structuring it “under the covers” (where programmers could see it), Metaweb lets partners and end-users create data structures or add information to structures created by others. For example, Metaweb/Freebase has an extensive structured database of movies, actors who appear in them, and their directors. You can ask (and get the answer) to “movies directed by Roman Polanski” and get only those movies – not those in which he only appeared. Try doing that with Google. You soon will be able to.

Other categories include business (with entities such as employers, industries, and employees), biology, space flight, and many more, and include representations – such as graphs, timelines, and tables – of how they are connected.

Most things don’t exist in isolation. They have complex relationships to other things, and by representing that information using verbs – for example, “the company that Google acquired” vs. “the company that Google competes with” – we can represent the world more accurately. And that means better, more meaningful responses when we search.