As I was reading Garet Rogers’s article, Google Image Swirl: Looks neat, but useless, I was struck initially by how strongly I disagreed.  But after spending a few minutes thinking about it, I realized we are looking at it from different directions.

Google Image Swirl is an experimental project focused on image categorization.  It really has nothing to do with "swirling" images around.  Google could have presented related images in some other UI, but the animated swirl is visually appealing and, yes, fun.  But that's not the point.

The point is that they are categorizing images based on image content.  Not just images that are related because they look similar--but because they represent ideas that are similar.

Imagine a photo of Mt. Rainier and a pencil sketch of Mt. Rainier.  If a computer could match those up as related images, that would be impressive.  Or a photo of Jay Leno and a hand-drawn caricature of Jay Leno... that would be impressive.

Google labs is a petri dish for innovation.  You shouldn’t look at experimental applications of new ideas and judge them against a mainstream application's business requirements. To do so is missing the point.

Years ago, my wife worked on a team at Bell Labs that built the first "object oriented" telephone switching system using an experimental programming language they called "C++".  That first system supported three telephone numbers.  That first prototype application would fail miserably as a commercial phone system.  But that wasn’t the point.

When an innovator looks at experimental research, she sees beyond the prototype to the possibilities the prototype inspires.