Document and image management systems have been doing this sort of thing for decades. Form a unique identifier for an image by hashing, split the bytes out into a path name by hex converting, and you have a file system path.

I'd say that this particular troll is going to get stepped on pretty hard, and all of these patents are going to be invalidated.

I never claimed I invented it ;-) It just seemed like the obvious way to prevent duplication when someone would rather update the same file again than search for a previously existing version.

In the end, when a dupe was uploaded for the second time, the uploader would be rewarded by not having to edit any metadata - as they would be redirected to the original asset page.

If someone ever uploads a second, valid, JPEG that collides in md5-space with a previously uploaded one, the sysadmins in the app group will receive an e-mail commemorating the fact someone defeated our deduper.

Does it know from which of a pool of servers should it get the data, based on that identifier? If not, it doesn't violate the patent.