http://groups.google.com/groups?q=freetella&hl=en&lr=&ie=UTF-
8&c2coff=1&safe=off&selm=Xns951A1E8DA9B51neo1061hotmailcom%4066.185.95
.104&rnum=2

It's a very, very, very bloody long posting, and the really relevant-
to-you stuff is only near the bottom, but there's a proposal for a 
file-sharing-and-searching-over-Freenet system, "freetella", 
basically gnutella + freenet as I understand it. The reputation 
management stuff is interesting (but full of *#&! math!) in 
particular, as well as the suggestion that Frost could form part of 
the back end of this thing too. The reputation management thing may 
be more interesting than Freetella proper, though, because it seems 
to offer what are described as, and I can't disprove it, "foolproof" 
ways to a) make any indexing system based on volunteer contributions 
to an index difficult to spam or flood or otherwise pollute based on 
reputation management -- I especially like the notion of contributer 
identities being in the key strings for their submissions so if they 
wind up in really bad repute clients can simply not retrieve such 
keys at all, thus allowing pollution to wither and die instead of 
persisting, but being ignored, but continuing to burden clients 
making sheer volume denial of service attacks of such publicly-
annotatable indices possible. Such DoS tricks will run up against a 
brick wall of clients not retrieving or propagating the pollution at 
all, limiting the damage to one node and its immediate environs. It's 
like they can dump toxic waste but it won't get into the groundwater. 
The math stuff has to do with preventing a spammer cheating the 
reputation management system by making multiple IDs that all vote 
each other up. It's tricky (Markov chains, what the [EMAIL PROTECTED] are Markov 
chains for Pete's sake?) and something is "an exercise left to the 
reader" but he (she?) seems to have found a way to make self-votes, 
even indirect ones, cancel out somehow. Also, there's no central 
stuff mentioned, which would have cast immediate doubt on the whole 
thing. No central reputation management, the reputation management 
reputation manages its own internal votes as well as whatever larger 
purpose it serves, no centralized anything as near as I can tell. No 
vulnerabilities. No dependence on non-Freenet services at all that I 
can see, aside from the inevitable loopback "spaghetti networking" 
internal to the node machines involved in everything Freenet, the 
basic internet protocols themselves, and DNS. Even that might be 
jettisoned when static IPs are used throughout (IPv6 might make 
dynamic IPs a bad memory. In a pig's eye.)

There's some sort of wacko anti-RIAA stuff in the article too, 
including an "I don't hate the RIAA, they're just a favorite whipping 
boy" sort of disclaimer right after a lengthy discussion of how to 
completely thwart the RIAA *without* freenet or heavy use of 
encryption. Whether we *should* thwart the RIAA is left as an 
exercise for the reader's consicence, I guess. The clever scheme 
involves breaking bootleg files up into chunks to small they are 
either unintelligible in isolation or fall under fair use, which need 
not even be encrypted though he seems to recommend some crypto, and 
are combined into the complete file (and if necessary decrypted) by 
someone who wants the file. The trick is for no host to offer more 
than one fragment so there's nothing but an unintelligible, 
suggestively named (author's own words more or less) or a fair-use 
quotation being shared, i.e. no probable cause for searching your cpu 
and finding the rest of the file or even supposing you knew what 
others were doing with the file. I'm not totally sure they wouldn't 
find some way to legislate such a thing out of the loophole 
described, though. In fact, they probably would. There's also some 
stuff about ISPs relaxing AUPs, freedom of speech, world peace, and 
so on. Says he's a fellow canadian -- probably a canuck that voted 
NDP on Monday and commutes regularly to an institute of higher 
learning then. And before that a long long list of suggestions for 
improving gnutella most of them seriously technical. I think he also 
took potshots at bill gates, shareaza, and some other prominent 
targets besides the RIAA. Oh, and the FBI and other law enforcement 
agencies of questionable trustworthiness. (Why not mention CSIS?)

Oh and he dares them to prosecute him under the DMCA for posting it. 
Then he thumbs his nose at the yankee gestapo and that's where he 
announces his canadianness. Hope he hasn't any travel plans to like 
Florida or Hawaii in the near future then. :)

Anyway the freenet-related stuff at the bottom looks interesting. I 
think the reputation stuff may be quite generalizable for a lot of 
other stuff. There's occasionally talk of how to influence unwanted 
stuff into expiring from the freenet here -- reputation management 
that blacklists keys (and bad blacklisters) in principle lets one 
stop their machine ever retrieving keys that 
are on a blacklist for something they don't want to make spread 
through 
freenet, e.g. child pornography. A popular blacklist could indeed 
depress the spread of a blacklisted file, perhaps to the point it 
can't be found in any data store but that of the one loser who keeps 
reinserting the thing.

_______________________________________________
Support mailing list
[EMAIL PROTECTED]
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:[EMAIL PROTECTED]

Reply via email to