--- Toad <[EMAIL PROTECTED]> wrote:
> On Sat, Nov 01, 2003 at 03:19:43PM -0800, pineapple
> wrote:
> > Uh oh!  I have another idea for this!  Another
> > weakness with this attack is the assumption that
> is
> > made about what constitutes a nearby key.  Right
> now
> > all nodes arrange their estimator keyspace
> internally
> > in the same manner.  Why?  The keys are encrypted
> so
> > they are essentially random and therefore they
> have no
> > relationship to each other based on their content.
>  My
> > proposal is to randomly scramble the estimator
> > keyspace using a function with a 1:1 mapping
> between
> > the normal ordering and the new ordering scheme (I
> > have no idea what function would be best for
> this). 
> > Now an attacker has no idea what key would be
> "near"
> > the target key because every node would have it's
> own
> > secret ordering for the estimator keyspace.  Two
> keys
> > may be near each other on one node but very far
> apart
> > on another.  Please keep in mind I'm talking about
> the
> > estimator keyspace only.
> 
> This destroys any chance of Freenet routing working.

I'm assuming you have been following this thread so I
won't repeat what I have typed earlier.  I'll try to
explain my thoughts on what happens to estimators
under my system.  Suppose we have 3 nodes in our
network; A, B and C.  Now assume that the estimator
graphs that A and B have for C look something like
these:

A's estimator graph for C (specialization is key x)
|___|___________|

B's estimator graph for C (specialization is key y)
|__________|____|

Ok, something strange here, A and B see C as
specializing in different areas of the keyspace.  This
is because A and B have used a reording function on
their estimator keyspace axis and so C looks different
to each node.  So, what does this mean?  What is C's
true specialization, is it x or y?  The answer is
both, and neither.  There is no "true" specialization,
any questions about specialzation has to be in the
context of the node that is estimating C.  This is
what nodes do now, they only make estimates from their
own context, doing otherwise would be a potential
source of abuse (with possible exception for new nodes
that don't have any routing info themselves).  I think
that something like this is happening on the network
right now, not due to keyspace reordering but because
the estimators themselves are subject to
network/cpu/storage/uptime factors.  I think this may
be the "non-obvious" specialization that Ian was
talking about.  The only way to find out would be to
gather the estimator graphs from many nodes and
compare them to see if a node has some kind of
"global" estimator graph, or if a node has a different
graph for each node that has an estimator for it.

> 
> -- 
> Matthew J Toseland - [EMAIL PROTECTED]
> Freenet Project Official Codemonkey -
> http://freenetproject.org/
> ICTHUS - Nothing is impossible. Our Boss says so.
> 

> ATTACHMENT part 1.2 application/pgp-signature
name=signature.asc
> _______________________________________________
> Devl mailing list
> [EMAIL PROTECTED]
>
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl


__________________________________
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
http://antispam.yahoo.com/whatsnewfree
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to