Tag Archives: ibm

What(son) Happened to Google

IBMs Watson computer wins Jeopardy tournament of champions.
IBM Watson\’s final score on Jeopardy.

Last year, Paul Graham wrote an essay about What Happened to Yahoo.

In summary, Yahoo was getting paid so much by their legacy business, enough that when technology moved on (faster than the market, as usual), they were incapable of changing (because they were getting paid too much not to change), and so they got left behind.

On February 16th, we saw that IBM’s Watson computer had defeated the two greatest human contestants to ever play the trivia/wordplay television game show Jeopardy!

Last night, it occurred to me that Google’s getting paid SO much by a market where humans do all the search. Will it recognize that technology has just (potentially) moved on?

The end of Google may not come through social networks (Facebook). We do plenty of stuff online that, for multiple reasons (not least to avoid boring our friends to death), we are not going to want/need to share with our friends.

But I do see a clear path from where we are now to a future where search robots are researching and organizing information for us. Our “virtual assistants” become truly virtual.

A lot has been written, including the Paul Graham article above, about Facebook’s “hacker culture.” Another way of looking at this, from an outcomes perspective, is to say that Facebook rewards people who build things even more than Google (and their vaunted “20% time,” which we never hear about anymore), and so for the moment, Facebook is winning.

This “corporate cambrian explosion” will likely be come to seen, or has already been seen, by someone on the HBS faculty, as a distinct and repeatable phase in the creation of incipient monopolies. There is a “golden age,” after exponential growth begins, but before the primary business model solidifies, when engineers are maximally rewarded for creative output. This is celebrated in Silicon Valley from Fairchild Semiconductor on down, but I bet you can find it in airlines, autos, oil, trains, every major non-military technology back to the dawn of the industrial revolution. This would be a great read…

I think it’s a fair generalization to say hackers have a low tolerance for things that bore them, they are intrinsically motivated. I had a co-worker once who needed to move, and he wrote a program to monitor all of the NYC property management websites for new no-fee listings. He reduced the time needed to research housing opportunities by a factor of 5-10 (at least compared to a non-coder like me).

Internet search is incredibly useful at the moment, but when the tools to automate information retrieval get good enough, we will move on to even more interesting things, just like my friend did, thanks to his apartment search robot.

A robot-optimized search engine might become a true competitor for Google. It might even be able to monetize the traffic through ads, exactly as Google does, if the ad seems like the best solution in a Watson-style analysis, the machines might be happy to explore sponsored links, just like we do.

There is no API for Google’s index. Google is researching machine learning to improve search results, which is conceptually similar to Watson’s information-retrieval AI, but the Google project, code-named “Seti” is trade-secret, you won’t find it hosted on code.google.com.

UPDATE 10/5/2013: Oh, look the punditry caught up with me, 2 years, 7 months, 14 days, 18 hours, and 50 minutes later. lol