First published July 19, 2007 in Mediapost’s Search Insider
For the first time in a long time, I’ve been consistently frustrated with the result that Google’s been returning for several of my searches. It’s not that Google’s getting worse, it’s that the nature of my searches has changed significantly. My searches are getting fuzzier as I’m stepping into territory I don’t know very well. Google is not functioning terribly well as my “discovery” engine.
Aaron’s Ambient Findability
Aaron Goldman wrote an absolutely fascinating column last week about ambient findability, based on Peter Morville’s book. I’ll definitely be taking Aaron’s advice and ordering my copy from Amazon soon. The interesting thing was that I read Aaron’s column shortly after I did an interview with Jakob Nielsen where he expressed similar cynicism about the practicality of search personalization. To sum up, both instances pointed to the fact that doing personalization is very difficult to do right. It’s probably impossible to do perfectly. But then again, personalization shouldn’t be perfect because humans aren’t. There will always be the human element of variability and unpredictability.
Google’s limits as a discovery engine
As much as the topic of ambient findability fascinates me (I explored the territory myself in a previous Search Insider ) I won’t steal Aaron’s thunder because I know he’s doing a follow-up column this week. I’ll take a more mundane path and talk about my increasing level of frustration with Google.
As I mentioned in last week’s column, I’m currently doing research for a book. Right now, what I’m researching is the nitty-gritty of why and how we make purchase decisions. By the way, Aaron suggested an interesting book, so I’ll do the same. Please do yourself a favor and pick up a copy of Clotaire Rapaille’s “The Culture Code .” This is one of the most fascinating marketing books I’ve read in some time. Rapaille talks about the challenge of doing traditional market research in trying to uncover people’s attitudes towards brands or other aspects of our culture, like food, healthcare and even the American presidency. The problem is that in most traditional market research vehicles (focus groups & surveys) we’re stuck with what people say. It’s almost impossible to uncover what people really feel. What people say comes directly from their cerebral cortex, the logical and rational part of their brain. But what they feel comes from the limbic and reptilian part of the brain, the dark, shadowy corners of our personas. The minute you ask them a question, no matter what the format, you immediately get the cortex in gear. This got me thinking about neural marketing and the actual mechanisms in our mind that click over when we make the decision to buy or not.
Rapaille’s book simply served to whet my appetite. I voraciously started looking for more of the same but books, research or articles that explore the primal reasons why we buy seem to be few and far between (hint: if you know of any, please pass them along in the Search Insider blog so we can all share). I turned to Google and tried a number of queries to try to dig up academic research or Web sites on the subject matter. I was definitely venturing into new territory and while Google usually acts as a reliable guide, it was leaving me stranded high and dry in these particular quests.
Personalization is an idea, not an algorithm
So, let’s get back to personalization. Would personalization in the form (Kamvar’s algorithm) that is currently being envisioned and rolled out by Google help me in this matter? Probably not. The signals (search and Web history) would be too few to help me zero in on the content I’m looking for. It wouldn’t really improve Google’s utility as a “discovery” engine. It would run into the same road blocks that Aaron and others consistently point out.
But here’s the thing. Google is making a huge bet on personalization. But personalization is not the only thing Google is working on. Personalization simply acts as a hub. MIT’s Technology Review recently did an interview with Peter Norvig, Google’s Director of Research. Norvig is, quite literally, a rocket scientist (he was head of computational sciences at NASA in a previous life) who is taking Google’s research in some interesting new directions. Speech recognition and machine translation are two notable areas. Speech recognition can overcome some major input obstacles not only on the desktop, but, more notably, on mobile devices and on a convergent home screen that fully integrates our online world and entertainment options. And machine translation can enable a number of automated systems that can power further online functionality. Both are very much aligned with Google’s engineering view of the universe, where introducing people into the equation just introduces friction in an otherwise perfect world.
But the really telling part of the interview came when the conversation turned to search. Norvig talks about the current imbalance of search, where there is an avalanche of data available but the only gate to that data is the few words the searcher chooses to share with the search engine. We’re trying to paint personalization into a corner based on Google’s current implementation of it. And that’s absolutely the wrong thing to do. Personalization is not a currently implemented algorithm, or even some future version of the same algorithm. It’s is an area of development that will encompass many new technologies, some of which are under development right now in some corner of Google’s labs.
Personalization, in its simplest form, is simply knowing more about you as an individual and using that knowledge to better connect you to content and functionality on the Web. There are many paths you can take to that same end goal. Sep Kamvar’s algorithm is just one of them. By the way, Norvig’s particular area of expertise is artificial intelligence. Let’s for an moment stop talking about personalization and start talking instead about what the inclusion of true artificial intelligence could do for the search experience. But artificial intelligence requires signals, and personalization is a good bet to provide those signals. It doesn’t have to get it perfect every time, it just has to make it better.
Just as a last point, Marissa Mayer said in an interview that Google’s current forays into personalization serve no other purpose than to give Kamvar some data to play with to improve his algorithms. We’ve all quickly jumped on personalization (and yes, I’m probably the most guilty of this) as the new direction of search, but many of us (and I believe my guilt ends here) are making the assumption that personalization means a form of what we’re seeing today. It doesn’t. Not by a long shot. And, at the end of the day, what we’re looking for is a jump ahead in matching our needs with what the Web has to offer. To win, Google doesn’t have to do it perfectly. It just has to do it better than everyone else.