"Mark J. Roberts" wrote:
> 
> I have no idea how the unrequest actually works. I'm under the
> impression that it works per-node and not per-reply, i.e., "The node who
> sent this response is evil! Punish it!"


Well, what if a given node provides both a good reply and a bad?  Does the node 
as a whole
get punished?  I was under the impression that unrequest applies directly to a 
given meta
data key.  After all, you have no idea (at least, not after the fact) which 
node uploaded
what meta data, and thus who the originator of this spammed data was.  All you 
have
currently is the meta data itself, and so you attempt to make this data 
'tagged' as spam,
via unrequests, so that it appears worthless (no one is matching this query, so 
it is not
relevant) and thus will be dropped out of the network and replaced with more 
releveant
data.

So this is where the unrequest for every peice of spam, or at least, the 
majority of it is
important.  Otherwise it will appear to be relevant, and propagate further when
overlooked.

So again, this assumes that each peer actually downloads the data associated 
with the meta
data key.

If that makes sense...  

At any rate, I do not fully understand it either.  I am still looking for Ian's 
proposal
document.




> But how that node is identified I do not know. In particular, I think my
> evil node could attribute the evil search results to an arbitrary node.
> 
> Maybe it's per-reference after all--"You shouldn't ever reply to
> searches about 'democracy' again, because you really fucked up this
> time!" But somehow this feels wrong...
>

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to