Ok, I just came up with this, and I've been up for quite awhile, so forgive me.
The way the RIAA sues you in court is by showing that they downloaded a song from you. So with that in mind.
A file-sharing program that scans the files you wish to share. Then googles a random set of small images
It takes those small images, looks at them binarily, ie. 011010010001101011.
It breaks one of your files down to its binary data.
It then scans all of the googled images, making note of long runs of binary data that coincide with the binary data in your file.
It then creates a list of image links ie. http://www.cnn.com/photos/f8a8.gif
at the end of the url it lists the segments of the image that it needs to put the file back together ie. 0x7623-0x7691
It will take ALOT of images, the url list will be HUGE (albeit VERY compressable), but in the end all you did was have a big file of picture links in form of a text file. NONE of the DATA that makes up the orignal file is coming from your PC. It distibutes the bandwidth across the internet.
The PC downloading the list goes and downloads the images, takes the segments out, peices them together, and out comes the file.
The list doesn't have to be in the form a "file", it could be done on-the-fly, just transmiting the urls and segments to the guest computer as requested.
A small degree of parity could be added at the end to insure missing or slow urls are able to be fixed.
Depending on how fast the computer and connection are, it might be able to stream it in real time from the urls list. (With a descent amount of buffering)
It's not a form of compression, in the end the url file may end up being much larger. It's not even a link to copyrighted material.(read .torrent) If a list of urls to images on the internet is copyright infringing, let me know. And google too.