Google Unveils Universal Search: Making A Hard Search Look Easy
While Google’s announcement of “universal search” is likely to get a lot of press today, my guess is most users won’t even notice it. That’s because the concept is so simple - that you get the results that are most relevant when you type a search in the main page. It even takes some explaining to remind people that search usually hasn’t already been this way. Until now, you basically got different results from a Web search (the default on Google and all the other major search engines) that you got when doing a search for news, images, books, maps, etc.
Now Google will integrate all these results in the main search results page, though you will be able to constrain your search to a particular type of search by clicking on a new navigation pane on the top of the pages. It sounds simple enough, but as Google VP for Search Products Marissa Meyer explained, “making Google easy is hard.” There was a reason that the searches were kept in different “information silos,” she explained - it’s hard to figure out what information to integrate so that you get very relevant results for the searches. For instance, if you type in the name of a person, you may want to see pictures, videos, etc; while if you’re looking for restaurants, you might want local information and maps.
On some queries, the result can be truly impressive: for instance, when you search for Martin Luther King’s “I have a dream speech” from YouTube and view it in place, right in the search results page. When I searched for Sergey Brin, the page started with pictures of the Google co-founder, then had Google desktop results from my computer, then typical web results, and finally news and blog posts about him.
Google will be rolling this out on its various servers over the next day or two, but you should try it yourself to see how you find the results. My guess is that unless you’re explicitly looking for it, the results will just seem better, but you won’t really notice the changes in the search engine itself unless you are explicitly looking for it.
That’s not to say everything is perfect. In fact, one of the themes of today’s “searchology” event was that “Search is Hard,” a phrase I heard from everyone from the first speaker, technology director Craig Silverstein to the last, Google founder Brin.
And there are some areas where I think Google’s results may not match all its competitors. For instance Google is indexing Google Video, YouTube, Metacafe and a few other video search sites, but don’t match the video search of a site like Blinkx. And for local data, I’ve often gotten better results from Yahoo!. But I do like the concept of “universal search” - aggreagating it all together - is an important one, and in doing this, Google retains its lead against the other search engines in terms of the general utility of its site.
Some of the other things Google showed today were extremely interesting as well, including two new views that initially are part of the Google Experimental page - a timeline view and a maps view. Both have interesting ways of displaying data. (And Google says it will now let you signup to have the “experimental” services part of your default searches, a nice way of getting more people to test the concepts.)
But more impressive was a technology called Cross Language Information Retrieval that was demonstrated by VP of engineering Udi Manber. Essentially, you choose an option called “search in other languages” and you can get back documents translated into another language. The best demo of this involved searching in Arabic and getting back documents that were originally in English, but translated into Arabic. At first, this will be most useful for people who are searching in languages that are not as well represented on the web, but I can see it being useful for English speakers as well. Google didn’t announce when this would launch, though they said it would be soon. Later, I asked Brin about translation technology and he indicated that he thought it wasn’t good enough for things like novels, it was already good enough for many business and personal information needs, and getting better all the time.
Eventually, of course, all these features may well just turn out to be part of the way we search. And then I expect users will take them for granted.