Google’s Incremental Search Results

by on August 29, 2010

New in Google’s live testing is what Jef Raskin described as “incremental search” (also jokingly referring to the dominant search pattern as “excremental search”) about 10 years ago. He predicted it would be usually the best way to perform free-text queries like this. At the time, few systems were really able to implement it, so it was hard to say for sure.

The first thing I’d say about Google’s design is to congratulate them on trying it, because there are obvious practical issues that might make the approach fall down. Most people tend to look at the keyboard first when typing long strings, which in turn removes much of the utility gained by each keystroke increment. They won’t see a lot of the results until they look back at the screen – by which time they’ve probably finished typing. The second issue is that the search results layout as Google currently have them are not very well arranged for incremental revelation. By having comparatively large amount of text that’s unrelated to the string being typed, they are designed for perusal after a search is triggered, not during its construction. There’s just too much to parse on the screen, changing extremely rapidly.

So I hope they will rapidly start experimenting with the layout and content of the search results when deployed incrementally. This is because I believe their approach is correct, but it would be a great shame to conclude from such an obviously flawed experiment that it was in fact the wrong one. Google’s implementation might be a very good example of how interaction design can be extremely hard to get right, and either works wonderfully or just sucks. In releasing this design, I assume that, like me, Google’s designers’ work is often blunted by the fiats of assorted suits and higher-ups who have reasons to look backwards when presented with new design ideas.

A final thought: even if the current design has flaws that limit its utility, the one thing that the likes of Google, Facebook, and others have in this is traffic and repetitive use. It would be wonderful to make incremental search the norm and to have people expect it. From the video you can see how incremental search really comes into its own by rapid iteration on the search string being used. Adopting such a practice might well produce a step-change in the efficiency with which ordinary people use free-text queries on large data sets.

Comments

Yes, albeit mostly in the form of “auto-suggest” and not really the full strength version described by Raskin. But still, happy days, eh?

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>