Google’s Hidden Gigantic Visual Repository

As I was reading through Searching for images with filetype: on Google? by Dan Russell of Google, I was not that surprised that filetype: takes different arguments and finds different things on Google.com and image search. I had seen this behavior. But this was stunning:

“to find an image, you have to use Images.Google.com. “

Dan’s post implied that

Google’s index for images is separate from its main index.

I emailed Dan and asked whether it was, indeed, true. Yes, he replied:

“Images, Videos, News, Web, Scholar (etc.) are all in separate indices.  Normally we blend all of the results together, which is why you’ll see images in a query.  BUT if you search “All” for image filetypes, you won’t find them because they’re in a separate index.  No blending is done with specialty operators (e.g., filetype: inurl: etc.)” 

I have observed some search strings returning no results on Google but some in Google images and was wondering why. That explains it.

Google’s images are on the surface web and can be easily discovered. But those that did not make it to the main index are rarely found because people do not search in Images.

The same is true about all other Google’s specialty searches (such as Google Scholar), as Dan confirms  – they, too, have separate indices.

Very interesting! There are implications for OSINT, sourcing, and any type of research. Live and learn.