I would vote for a preview funtion, as well as fixing the issue with "already on list" not being updatable.(see my other post on this)

next would be the multi temp issue, where several temps can contain excatly the same file.

XLX can download the same file, from different hosts, into several different files. this slows down a fast process. Why not download the same file, with several names and from several computers, into one file?

and lastly is a big step. Why not find the files that are most commonly being passed, and store parts of them in open space, in spare time.

the more you spread the information, the more people can acess it quicker.

This is a good reason to reach into the "duplicate recognition" areas, since it would have all kinds of issues of recognizing what part of what file on what computers.

I guess some may have a problem with being forced not to freeload, though.