Thoughts on Google's latest search announcements

I spent a good portion of Monday in the company of Google, as it unveiled a series of announcements about search technologies - such as the inclusion of new "real-time" results culled from Twitter, Facebook and the news media and planted, dynamically, into the results page.

Given the obsession with the real-time web at the moment, it was obvious that most of the attention would be focused on a sort of OMG TWITTERZ IN MY GOOGLE!!! reaction, but I was struck by a number of things that seemed highly significant. Here they are:

-Google's user interface is getting more cluttered all the time. Yes, the famous home page may be getting lighter, but that doesn't seem to apply to the layers of pages it takes you to. Finding real-time search results requires a series of clicks, none of which are blindingly obvious or quick for new users to get to. It's time for Google to start applying its reductionist approach more widely.

This is a way to move into highly personalised search. Once you've got users looking at social networking results in their feed, it's only a matter of time before they start demanding more access to the results that matter to them. Google's always been slightly offish about giving answers based on your personal connections - it loves the feel of a cold, hard algorithm - but social search is a powerful idea. And it's worth knowing that Bradley Horowitz - who outlined his ideas on social search to me four years ago while he was working as a senior executive at Yahoo - is now Google's vice president of product management.

- Google's mobile search is getting very, very strong. Even with the Android operating system and other developments, I think people underestimate how much mobile phones matter to Google. To me, it's pretty clear that they think mobile will be the area where they will develop their next big cash cow. The company is currently working with two major US retailers, Best Buy and Sear's, to include stock/inventory listings in mobile searches. The idea is that if you search for a product on your mobile phone, it will be able to hook into their systems and tell you if there are any items in stock nearby. That's a big money-making opportunity. Really big.

- What is the possibility for Google Goggles? Computer vision has been the holy grail for all number of technologists over the years, and if Google throws its significant resources behind the effort, it could get somewhere. They could give it a better name, though.

In fact, in the process of asking a couple of questions, I pushed an idea that crossed my mind - focused on who "owns" the Goggles references (the example, who decides that a picture of the Empire State Building is actually the Empire State Building). Vic Gundotra didn't answer my question, but passed it over to Hartmut Neven (the founder of Neven Vision, a company Google acquired in 2006 and the developer of the system).

Neven said the canonical database of references was constructed algorithmically - unsupervised learning, he called it (it sounded quite a lot like Google's general search system). If that's the case, then, does that mean that the image of something could be googlebombed?. "Theoretically, yes," he said - but, he added, they thought they had systems in place to prevent that.

What an idea. Computer vision is amazing, and if it can be applied to search then just imagine how powerful that could be. But, equally, imagine a situation in which, say, a product is googlebombed by its detractors - or even a competitor. Suddenly the idea of "image rights", something usually associated with celebrities, becomes absolutely huge.

I'm interested to see where this all leads.

0 Comment



Captcha image