May 19, 2008
Google and the heavy lifting of search
There was a distinct lack of fireworks at the ”factory tour” that Google put on today to show off its search prowess. More than anything else, what came across (and what always comes across when Google engineers take the stage) is that making search work really well is difficult.
Most of the time was taken up with incremental improvements in two of the most promising areas: local search and “universal” search. These are not new, but they have the potential to greatly enhance the relevance and range of material returned by regular Web searches.
When I had a moment to talk to Marissa Mayer, vice president of search products and user experience, she concurred that most advances in quality are largely invisible to the average user, though an occasional eye-grabbing feature - such as the video links returned in universal search results - are more noticeable.
According to Mayer, around 30 per cent of Google’s search results presently fall into the “universal” category (meaning that they draw on a search of seven different sources - videos, maps, news, books, images, products and blogs.) That proportion will grow as Google works out how to return relevant blended results from these different sources for more types of query, she added.
Universal search has been talked of as the Holy Grail of the business, drawing on all types and sources of information to deliver the most relevant result. Yet to judge from what Mayer says, the real “killer app” of universal search has yet to be seen. This will come from extending search to social networks and drawing in information that is deeply personal to the user and their network of friends and contacts.
Footnote: What fireworks there were today came with the formal launch of Google Health, a service for storing personal health records online (limited for now to the US.) This is a project that has attracted an inordinate amount of attention in the time it has been under development, but it is still hard to see it as a mainstream service for some time. Few health records exist in digital form, and while it sounds a nice idea in principle, there seem to be few incentives for now for most doctors or patients to change their behaviour.
