Saturday, September 20, 2008

Google's strategy - connecting the dots - I

Many of us have started using Google's new browser Chrome, being impressed by some of its features (which I still believe have been lifted from its rival browsers). But, what exactly is the story behind its investments into this already (seemingly) stable segment of browsers. I have been thinking of Google's strategic motives for long, and with the launch of new products and talks of few more, things are getting clearer. Let me explain.

Start with its mission "to organize the world's information and make it universally accessible and useful". Most of us wrote it off as simply a way to say 'better search'. But they have been restating with every product launch of getting closer to this mission. With Marissa Mayer ans Susan Wojcicki posting constantly on their blog, about the next 10 years for Google, its obvious that they are not sitting quiet. In her post, Marissa speaks about how she would want the search engines to work upon the modes of search, media covered under them, personalization and language features to give a better user experience. When Google started with its GOOG-411, it was seen as a voice enabled search. But later they accepted that its a way to train its audio engines to recognize text withing audio. That learning may even have been incorporated in its GAudi, audio indexing feature. Now, it has added features for you to overlap captions on Youtube videos. I'm sure this is again a way to train its video engines, to 'learn'. One of the features that I was expecting Google to launch was face recognition. And here it is. This provides Google the power to train its algorithm to perfection (Anyone working on face recognition would agree that the larger the face base and the manual confirmation, the better would be the accuracy). Basically, it is employing YOU to check whether the algorithm is working fine or not. Not that their employees aren't trying. One of its patents talks about recognizing text directly in images, be it computer-generated or photographs (captured by its cameras for Street View) or a series of photographs (videos). Which means - given either an image, video or audio - and it has a way to find out information from it. Picture all of this put together. You give a search term,e.g.MacDonalds Mumbai. The search engine, of course can give you results from conventional text web(not only English language. Google is improving its translation tool to expand the number of languages that it can search, so pages of most of the languages are still available to you). It can also do it for any piece of audio (probably a jingle), or video (any MacD ad) which mentions the phrase. Any image on a website or photograph from someone's public album or an photo present in Google Maps(an image of the outlet) which contains that text is returned to you. For that matter, why should you even give a term. You could even give it some random image of a person whom you would wan't to find out (like the ones that we generally need in quizzes & puzzles) or some piece of audio, that you would want to know more about. I even believe that Google would have been working on the other facets of an audio, like distinguishing voices or finding out genres. May be useful, when you would want to find out who was the speaker in some speech that you have been hearing (or in the Indian context, which raaga does a particular Carnatic song belongs to) - it's nothing but identifying patterns. Isn't all this 'organizing world's information'? (Of course, analysts also spoke about extending it to barcodes, RFIDs and ISBNs and letting you know which product is available in which shops)

With its development in the mobile space, and providing access to all the features on the modern-phones, it is making sure that you need not wait till you reach your computer or internet to get the information. With its Android powered phone, Dream, launching just few days from now, it would want a better control of this all-important medium. Talking of personalization, there has already been a big hue and cry over the information that Google stores about its users. Using your past searches to find out whether you are wildlife enthusiast or a golf follower, would help its search engine to present appropriate results when you search for 'tiger US'. Similarly, given the location, especially, relevant for your mobile phone, it can provide you the local info accurately; when you say MacD, it would make sure that nearby outlets are presented to you. Isn't this "making info universally accessible and useful"?

This is about some of the obvious steps that it has taken in its homeground-search. For restraining the length of this post, I would be going ahead with its other revenue-centres in the next post.

No comments: