Message routing is not necessarily contributing to the network, it always depends on how many connections you have, and I heard there were people who get a couple of hundred results and drop all connections to enhance their downloading experience...

Yet, I don't think there are too many freeloaders using gnutella, since there are networks where freeloading is made much more enjoyable, because they have a lot more users and ten times the content of Gnutella.

Btw: many of the assumed advantages of swarming are not all quite proven yet and I doubt freenet-style swarming will still be possible AFTER introducing supernodes. A shielded client would no more see, which files are requested most often, so he wouldn't know what to download. And it could create a lot of traffic... Is anyone able to describe how it should be implemented? E.g. how big should the partial files be, how they will be identified and in what way a client could decide which files to download? The most interesting question would be, if this will be efficient, since Gnutella, as a chaotic network, organizes itself to a certain extend, and often requested files will be shared more often automatically after a certain time...