The important distinction you must make here is that the actual act of downloading the random bits of data is not copyright infringement. In the OFF system, pieces are COMPLETELY randomized, and do not represent the works which are being derived. Not only that, but also pieces can be reused. For example, if I created some blocks to store an mp3 I recorded, those same blocks could potentially be used to recreate a home-made video which someone else stored on the network.

Now on the FLIPSIDE, you are correct. The actual act of putting together the pieces to create a copyrighted work is, in itself, infringement. But in this system, you can legally share the pieces without fear of persecution, because without the URLs, the pieces are meaningless.

So now instead of on the host's side, the legal question lies in the user's hands. If the user chooses to assemble the correct blocks using a URL which clearly leads to material that is copyrighted, they are making a conscious choice to break the law.

There are many other legitmate uses for this peer to peer system as well. Some additional benefits of this system are the distribution, and the fact that blocks can be reused. If a file is successfully propogated in the system, you could potentially have a free alternative to paid file hosting, which sounds like a bargain to me.