Google recently introducted another little browser toolbar widget called Web Accelerator. When you go to a page, the Web Accelerator will look at links on the page and, under certain conditions, start downloading the linked pages automatically in the background. Then, if you click on a link, the page and images will already be downloaded so you don’t have to wait.
Aside from the havoc wrought with web server statistics, an already murky area of accuracy, it sounds great. Unfortunately, it looks like whatever Google engineer was working on this ignored the realities of modern web development. This 37 Signals: Signal vs. Noise thread has a good wrap-up of things as well as a work around or two, and this site has some brute force instructions for blocking the Web Accelerator from your site entirely. This “Backpack” page seems to be keeping track of different ways to work around the issue.
If you’re in PHP land, you can check for the existence of the “HTTP_X_MOZ” header with the $_SERVER[‘HTTP_X_MOZ’] environmental variable, although it’s not clear to me at the moment if Web Accelerator uses this header or if it’s a Mozilla specific thing. The above backpack page seems to think it�s there, but I haven’t had a chance to check things out myself.