there's a lot of confusion here, but for the record as I understand it ...

Swarming: A file is uploaded to the "network" and stored on any number of different peers in bits and pieces. The idea is to have more availability and redundancy. As far as I know only Mojo and Freenet do this. Also IMHO it doesn't work as good as it sounds - with big files you eventually wind up with pieces missing making the whole file useless.

Segmented Downloads: This is simply downloading parts of the file simultaneously from different people sharing it. It works well as long as the file is being shared by more than one person. Both FastTrack and Gnutella have the possibility of doing this although only a few Gnutella peers currently do.

One big disadvantage Gntuella faces is it must rely on filename or keyword searches to find more sources. This isn't too reliable and each source found must be checked by the peer before it can be sure it's really part of the same file. FastTarck can search based on the file's hash alone and reliably find other sources even if they're named completely different.

One of the problems with LimeWire's "UltraMegaSuperPeers" (or whatever they're calling 'em these days), is it still doesn't seems to provide for reliable location of additional sources. In fact, finding more sources - even just a single one to continue a broken download - has always been one of LimeWire's major weaknesses. The developers have repeatedly claimed LimeWire resumes downloads, but any user can tell you it's rare that it works, and once a file goes into the "can't be downloaded" status you're SOL. Very bad if you have 90meg of a 100meg file! FastTrack always keeps an eye out and will pick up new sources when they come online. Given enough time, FastTrack will eventually get all of even the biggest files.

What makes me frustrated is LimeWire had the potential to match this but seems to have flubbed it. They had to redo the protocol for SuperPeers anyway, so why not do it completely right this time (or at least match FastTrack)?

Swarming is ok, we'll probably move towards stuff like that in the future, but right now there's still some more important improvements to be made to the file sharing system.

Segmented downloading is really working great on gnutella, very versitile, files of the same size in bytes are grouped together and used as sources for a single download. Corrupt chunks are filtered out by using roll-back verification, which compares the overlapping ends of chunks to make sure they contain the same data. In real world distribution this is working 100% of the time, and corrupt chunks rarely even come up. Good news, but there's more to be done.

BearShare has contributed the proposals to search and share with file hashes. Also another cool contribution by BearShare is the X-Alternate-Location header, which comes up when you start downloading from a host, it will give you a list of other hosts with the same file. Often these hosts are beyond your viewable horizon, very useful feature.

UltraPeers proposal by LimeWire is cool also, Gnucleus and BearShare will support it soon. Basically it should expand the viewable network by a factor of 20 or more if all hosts are using it, also it will lessen the bandwidth needed by nodes to be part of the network.

The best of gnutella is still ahead.

in answer to the original post - ie why isnt bearshare / limeware / gnutella network as popular as morpheous the people i have spoken to who use morpheous have not heard of this network or those that have said "but its got spyware".

i doubt if any of these people know anything at all about swarming or fast track (i didnt even untill a while ago) and therefore think this is issue is irelavent to the original question.

i use winmx and gnucleus. neither of wich have got adverts. gnucleus so far wins hands down on the software side. yet winmx seems to have so many more of the files i am after but im
just not prepared to sit there and wet nurse it for 24 hours when trying to dl a large file becuase it just gives up so easily.

maybe gnucleus could become multi network?

Colin.

I think it's just that FastTrack works better right now. People notice they get better results - they really don't have to understand why.

Gnutella was first, and FastTrack was designed to account for many of its weaknesses. The only problem is it's not an open protocol, and now that they use a central server for authorization, it can be shut down like any other private P2P network. I just hope Gnutella catches up to it in functionality before that happens.

Of couse people could start implementing gIFT based peers using the open version of FastTrack. True, it can no longer work with actual FT servers, but you could form a new, more functional, P2P network around the protocol.

I agree, fasttrack has some pretty good advantages, but over time, the competition between bearshare, limewire, gnucleus and others is slowly making gnutella more powerful.

Limewire superpeers will open up the visible horizon on gnutella by a factor of 20 or more.

BearShare ping/pong scheme reduces traffic over 70%, opening up the network to better query and query result flow.

As clients implement these new features and others like hashing, etc.. gnutella is not a static protocol, every day improvements are being made, new proposals debated, and code written to bring it all together.