Google has a new images search engine, which is pretty impressive. Type in tiger and you get pages of thumbnailed images of tigers. Click an image and the original page loads, as framed by the Google result. At first, I thought the search just looked at the file name, e.g.,
tiger.gif, but then I did an ego surf.
The search for craig saila pulled up three interesting (and accurate) results using a smarter rules.
The first and second images have craig and saila mentioned on the pages, but the third seems only to have similar letters to saila in the name of the page the image is on,
www.canoe.ca/AirMergers/jan10_twaamranalysis-ap.html. What makes the third result so bizarre is the fact the image is one I helped create while working at CANOE Money.
At first, I thought Google had found some incredible search algorithm that realized I’d worked for CANOE Money, and that this image was created when I worked there. I quickly sobered up when I realized:
- the page was created after I left
- and Google didn’t even find my glorious head shot. (After writing this I decided to update my “robots.txt” file to slow the spambot tide and noticed Google’s spider was actually blocked from finding that head shot. In fact, the only images it displayed were those not excluded via “robots.txt.” Just goes to show the value of a decent “robots.txt” file. Just hope the spambots behave as nicely Google’s spider did!)
The fact Google didn’t display my head shot is an example of how the search is not quite perfect. The head shot appears on both my resume and bio pages, where my name is often mentioned. Despite that, the heuristics on this are pretty impressive, and it could make for a powerful tool..
However, when combined with Microsoft’s new Image Toolbar in IE6, Google’s Image Search makes it even easier for people to steal images from the Web. While it’s possible to remove copyrighted images from the results, it’s a labour intensive job, especially those site’s with poor architecture. (Prevent IE6 from displaying the image toolbar is even more difficult.)
If Google does indeed plan on going public, they could also make a killing charging for searches with the adult content filter turned off.