Good to hear from you again bigearedone
re the incompletes being able to be refound, I wish it were easy.
There does exist a unique identifier, called a hash (sha-1). It's used for the bitzi lookup, magnet links (control-click a file in your Library Pane to see one of those), to verify that the file you received *exactly* matches the one sent, and probably more. There is also a more recent method of checking smaller chunks of a file, generally callled THEX, and it is still being improved/implemented more widely. I asked zab (now HE is expert on this part) if maybe one could resurrect an old incomplete file just from the bits already downloaded, but it doesn't look likely. And so, if the data in the downloads.dat file becomes corrupt or useless, there's not a method currently used to rebuild it.
Also, searches by hash have been . . .in short, banned. Not sure of the exact reasoning, but they were found to be easily abused and the network apparently suffered. Hashes are also "expensive" in terms of CPU use, but I gather this is a tradeoff to prevent a network being flooded by fake files because of a quicker and less secure hash method.
Keyword searches will indeed find current sources, but the problem is the cost of matching the new bits with the old. Too, files which may seem similar (size and much of the name) may actually be different in a few bytes. I guess the answer is currently coded as "just start from scratch after 6 failed attempts."
btw--I've quickly found "reading the manual" is not very reliable, since the developers are rightly spending their time developing. heh--in a dynamic network, it makes sense that the info is dynmaic too: static manuals get outdated pretty fast. The new wiki at
http://www.the-gdf.org/wiki/index.php?title=Main_Page should help the developers update the info more, and I'm finding reading there quite interesting.
Sorry I can't be clearer with answers to your questions: I am not an expert, but you're welcome to what I've read
And yes, there is hope about the new spam. The LW folks are talking with the Credence developers, for example. I guess the risk is that any "banning" mechanism can be tricked to ban legitimate hosts, which poses a risk to the network. The more I read, the more aware I've become of the 'politics' of p2p network development. LW is getting to be quite the grandfather here with its upcoming 5 years of development, and its growth is more fun to watch than many sports matches