> > No, because nodes are free to alter HTL arbitrarily.  This is one of the
> > strengths of freenet.
>
> sure, i understand. but i guess that actually nobody modifies the HTL as
> such pratice may allow infinite-time-long searches if the document is
> not found!
>
> > Since it's impossible to prevent hostile nodes from
> > messing with HTL, freenet doesn't call this an attack.  Instead, it
> > documents it as a permissible part of the protocol, making proofs of
> > identity much more difficult.
>
> well, my reviewed question is: is that true that if your node recieve a
> [MAX_HTL] request it can be reasonably (80% or more) sure that the
> neightbour is the initial requester?

As far as I understand, absolutely no. First of all, I guess many people do
modify their HTLs - even if to some "popular" values like 15, 25, 50, etc.
Second, the way HTL works (or, at least, as it was supposed to work
according to original design), is that each node randomly either decrements
it by 1, or passes it unaltered, to the next node. This ensures reasonably
low probability of requests with very high HTLs, and at the same time gives
a very good probability that the requesting node isn't the first one in a
chain (this balance could be changed by choosing an approprioate probability
of decrement).

With best regards,
Victor Denisov,
CEO, Jera Systems (http://www.jera.ru)
Researcher, Moscow State University, dept. of CM&C (http://oit.cmc.msu.ru)


_______________________________________________
Tech mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/tech

Reply via email to