April 28th, 2008 by Fred
While we’ve remarked before that Google is blind, it turns out that the search behemoth is not satisfied with this handicap and foresees a future where the engine is able to algorithmically ascertain an images identity based on similar images in its database. Freaky, huh?
Search Engine Journal covered the story, which links up to a TechCrunch article with both more information on the research as well as the originating PDF.
While there are a few jabs at how “SEOs” will need to begin optimizing graphically optimizing images, the further implications are really much more sophisticated than what we expect from traditional search. After all, it’s only a few leaps from being able to algorithmically understand images to being able to do the same for videos, and tapping into the vast resources of images on the web that are currently NOT optimized for search opens up some really interesting doors.
Does it mean anything to you now? Not really. Just prepare to be embarrassed when your alt text bears no resemblance to the image it portrays… and Google gets smart enough to catch you!
