Those are some good points I didnt think about (I am definetly no begining user).

Right now its for the elite (people who have at least tried a few clients before settling with phex).

I think it very well could be made for the masses, but it would have to be packaged up for easy install on multiple platforms, and have fuller documention (especially for the newbies who have never used gnutella before).

I'd say a lot of users stick with Bearshare or Limewire because they are popular and they dont see a need to change, and are not actively looking for new clients. I for one originally used Limewire and recommended it to friends, then I tried BearShare and it seemed to be nice (I really didnt care about the spyware), but it would always reset my download list and there wasnt download grouping, Limewire on the other hand seemed slower or like there was something else wrong with it, that is when I started doing research in programming my own Gnutella client, I found a java library called Jtella, and thought that was a good route to go but it turned out to be an out dated library (it wouldnt compile or function), so I then remember source forge and did a search in there and came across phex, from then on I since havent used Limewire or bearshare, it did crash on me a few times which appeared to be random, and they were pretty severe crashes taking down my network card having to reboot to get it back, but that was really rare, and at the time I was experimenting and switching ports and connect counts when phex was connected to the gnutella network . But anyway I just saw my downloads increase by a lot, even a friend I recommended it to had also noticed considerable increase in the chances of downloading a file over night. The autosearch feature allows it to stay up to date on the changes in the gnutella network, and I can customize the search string for it.

I agree with that on the basis that some users will want a nice and clean interface (form over function), while others will want more functionality.

There is a few things I dont get though, why do you need to talk on the GDF to implement parallel downloading of the same file? I can understand that some extra protocols can be done to help improve it, but I dont see it as totally necesary. Its no diffrent then downloading diffrent files at the same time... The way GetRight handles it is by breaking the file into multiple pieces and then downloading the diffrent piecies from diffrent hosts, extra protocols can be used to prevent you from losing your seat (so to speak) while switching a host from one piece to downloading another piece, and at the end of the download the peices are recombined. Anyway, I will more then likely join the GDF and discuss this there.

I think another aspect that has not been mentioned is possibly looking into revenue models. Its cool that this is being done for free and open source, but most projects these days are looking into ways of getting money in. Freenet for example uses donations through paypal, and with this money they were able to hire a full time programmer, and it can be used to get publicity, this way Phex can get the recognition it deservers (and I really think it does deserver a lot of recognition it is IMHO a great if not the best way to use gnutella).