Inside Google's sci-fi search experiments

The familiar whistle of a smartphone notification told me that Paco was hungry. The dog's face had appeared on my screen, as it did eight or so times a day, to remind me to fill out a form. And I had an interest in feeding the beast -- my answers would help set the course for Google's evolution as a mobile search company.

Paco isn't a real dog, of course. The name is short for Personal Analytics Companion. It's an app that Google's user experience researchers rely on for their many studies about the search giant's usability. And what was Paco hungry for? Data. Along with 100 or so others, I was invited to participate in an October study that pointed toward the future of search. The study questions, coupled with interviews with Google executives, paint a picture of a company methodically building a search engine that would be at home in science fiction. And it's a search engine increasingly focused on smartphones and tablets.

Google has a strong incentive to focus on the mobile world. This week, for the first time since an analyst began tracking the number in 2006, desktop search declined. Meanwhile, search on smartphones and tablets is surging, and could account for a third of all search traffic by the year's end.

Against that backdrop, Google has faced repeated warnings that it could be ripe for disruption. The introduction last year of Siri, Apple's voice-powered assistant, led some to call it a competitive threat. (Google Executive Chairman Eric Schmidt was among them; never mind that Siri defaults to using Google search.) More recently, some have warned that Facebook's search ambitions threaten Google's dominance.

In response to this week's desktop search decline, Business Insider -- in a supremely Business Insider kind of story -- announced that we had hit "Peak Search," and that "the Google era may be over."

All these competitive analyses share the assumption that Google, surrounded on all sides by deep-pocketed competitors, will leave the search product as it is. In fact, Google is very much aware of the shift to mobile devices. Inside the Googleplex, some of the world's most talented engineers are working to ensure that the company dominates mobile search the same way it's dominated the desktop for more than a decade.

Spending time with Paco showed me some of the ways they're trying to do it.

The faster-horses problem
Whenever Paco whistled at me during the survey, which was called "Daily Information Needs," I opened up the app and filled out a form. It asked me to do several things:

• Name something I had recently needed to know, like an address, the name of a restaurant, or something more complicated.
• Describe why I wanted the information, how soon I needed it, and whether I needed it for myself or someone else.
• Say whether I had used one or more Google products to get the information.
• Describe my satisfaction with the answers I had found.

At the end of the day, I visited a Web site to give Google additional information about my searches and my satisfaction with the answers I'd received. The experiment lasted for three days.

On its face, the questions seem mundane. But they hint at one of Google's biggest problems in building a better search engine: with few exceptions, it can answer only the questions it's asked. People rely on it to retrieve information on all sorts of subjects, of course. But there remains a whole universe of questions that nobody asks Google, because nobody expects a search engine to be able to answer them.

Queries like: Show me an outdoor Lady Gaga concert in a place with warm weather where the tickets are under $75. Find me a martial-arts class that's no more than $35 that's close enough that I can drive there with the gas that's already in my car. What color should I paint my house?

"With Google search, we get a lot of different queries, and we can see the kinds of things people are using Google to search for," said Jon Wiley, the lead designer for search at Google. "What's more difficult to see are more overall questions people might have out and about in the world -- that they might not think to ask a search engine, or that it might be difficult for a search engine to answer."

One way of solving that problem might be to simply ask Google users what they'd like it to do. But then Google would likely run into the problem attributed (perhaps wrongly) to automaker Henry Ford: "If I had asked people what they wanted, they would have said faster horses."

And so instead, Google is asking users about their broader needs. What kinds of things do you need to know during the day? From their answers, Google engineers can devise new solutions. And in doing so, they move us closer to the all-knowing "Star Trek" computer that search chief Amit Singhal cites as his primary inspiration.

"It's one of the most informative things ever, to see people use your product," Wiley said. "There's a richness of information there that is really important to our work. A lot of what we do as designers is about empathy -- it's about putting ourselves in the shoes of others."

More and more, the shoes of those Google users are away from their desktops. Mobile search generates lots more queries for local data: restaurant names, nearby bars, movie showtimes, turn-by-turn directions. Queries also tend to be about more immediate information needs; users are less likely to leisurely research an upcoming trip on a smartphone than they are to search for and book a hotel room for the evening in a country they've just landed in.

Interestingly, though, Google says mobile queries aren't wildly different from their desktop counterparts. Many questions people used to ask Google before they left their desks they now wait to ask until they're in their cars. (Travel directions are a good example.) The implication here is that Google has less to lose in the shift to mobile than its competitors might hope. It has already been answering questions like this for years; the only thing that's changed is the device its customers are using to ask them.

Going beyond the box
Ultimately, the shift to mobile may matter less to the future of Google search than two other changes the company made this year.

The first came in May with the launch of the Knowledge Graph, which aspires to map every possible connection between people, places, and things. Today searching for an actor's name within Google now brings up a short biography, along with a filmography and related actors. That's the Knowledge Graph at work, assembling data from a variety of sources and mapping the connections between related searches.

It's also how Google can begin to understand queries when a single string of letters describes a variety of things. "Kiwi" can mean a fruit, a bird, or the people of New Zealand -- Google didn't always understand that. It does now. And it knows the connections each of those kiwis has to millions of other things.

"The idea of a database of everything -- it seems almost crazy that we could try to do it," said Jack Menzel, director of product management for search. "But technology has come to such a point that you can start to do these kinds of things. We've tried these types of projects over and over and over again. And we're finally coming to the point where modeling everything is actually becoming kind of feasible."

The second major shift, introduced at the Google I/O developer conference in June, was Google Now. Since Google first launched, using the search engine meant entering a query, either via keyboard or voice. Google Now threw all that away. The feature, which is available only on phones and tablets running the latest version of Android, tries to anticipate users' needs without them having to search at all. When I open it up on my Nexus 7 tablet, it shows me a whimsical drawing of San Francisco (where I live), the local weather, and the results of last night's baseball games in the Bay Area. (It knows I care about baseball because I recently searched for a San Francisco Giants score.)

Google Now couldn't fully replace the core search engine. But it points to a future in which people will use Google services more without feeling like they're using Google at all. Answers will materialize before they even ask.

"It's always about speeding things up," Wiley said. "How do we make it easier for people to ask questions and get their answers quickly so they can get on with their lives, and get on with the real work -- making dinner, picking up the kids, and all that stuff."

A continuing search
Spend enough time talking to Google search executives and it's hard not to feel that would-be disruptors should worry about Google, and not the other way around. Siri is still struggling to understand basic voice input, and Facebook search won't arrive for months at the soonest. Meanwhile Googlers watch sci-fi movies for inspiration and make gradual progress on turning them into reality.

"This isn't a one-off event for us," Wiley said, as the study I participated in came to a close, and his colleagues began to compile the results. "This is an ongoing activity."