Bailout

The power and serendipity of Google Instant

By Felix Salmon

Tyler Cowen has yet to make up his mind about Google Instant: it’s fun, he says, “but I find it distracting and it will lose me time in serendipitous diversions”.

My initial reaction to that is one of incredulity: Tyler’s a blogger, which means that he more than most people gets real, immediate benefit from serendipitous diversions. More generally, while I’ve never been particularly persuaded by the complaint that searching on Google involves less serendipity than browsing in a physical library, it’s pretty obviously no bad thing if Google starts throwing up serendipitous-yet-germane results in real time. Think of it like the random mutations which power evolution.

Google’s Matt Cutts explains how this kind of thing can be extremely powerful:

I was recently researching a congressperson. With Google Instant, it was more visible to me that this congressperson had proposed an energy plan, so I refined my search to learn more, and quickly found myself reading a post on the congressperson?s blog that had been on page 2 of the search results.

Ben Gomes mentioned this during the Q&A, but with Google Instant I find myself digging into a query more. Take a query like [roth ira v]. That brings up Autocomplete suggestions like [roth ira vs traditional ira], [roth ira vanguard], and [roth ira vs 401k]. Suddenly I?m able to explore those queries more just by pressing the up/down arrow key. I can get a preview of what the results will be, add or subtract words to modify my query, and hit enter at any time… When I was in grad school, I had a professor who mentioned that peoples? information need often change over the course of a search session. Google Instant makes that process even easier: people can dig into a topic and find out new areas to explore with very little work.

So I think that Alexis Madrigal is worried about nothing here:

I worry that Google is driving more traffic to the most statistically probable searches. The most-trafficked ways of searching for something will get more trafficked. I wouldn’t be surprised to see the number of unique searches drop because people see something in the list that makes sense, even if it’s not exactly how they’d have put it.

This may only be a slight narrowing of our collective imagination, but it’s worth noting because it’s another way in which algorithmic suggestions or restrictions shape our behavior, even (or especially) when they are soft and/or useful.

It would be interesting to see Google quantify this effect: I’m sure they have the ability to do that. But my gut feeling is that, contra Madrigal, the most-trafficked ways of searching for something will actually get less trafficked. Lots of people are currently searching for that congressperson, or doing a general search on Roth IRA. Those searches are now going to get much more granular and useful. The tail might get a little bit thinner, although I doubt it. But the head is going to get much thinner. Which will have an overall positive effect when it comes to shaping our behavior: we’re going to be much less likely to just end up going to the sites at the top of the search results for the most popular terms. And that’s got to be a good thing.

Full post as published by Macroprudential on September 08, 2010 (boomark / email).