Matthew Toseland wrote:
On Tue, Jul 12, 2005 at 03:19:21PM -0700, Tom Kaitchuck wrote:
Matthew Toseland wrote:
The network can't split. 50% isn't enough. You'd need a supermajority,
as I have explained - AT EACH NODE. It's not a global vote. Each node
would need 2/3rds of its connections to vote, and would need them
(including a biased score from the first layer of indirectly connected
nodes) to vote yes at say 2/3rds (supermajority depends on sanctions to
be taken, but 2/3rds sounds reasonable to trace the author). We would
probably have a large number of blocks to trace, so we could to some
extent work around uncooperative bits of the network, but the objective
is to find the node that posted the data, or the sub-network that it is
part of. We can then either warn them, or sever connections with them
(either all connections or premix connections).
I was assuming that there would be enough vairation in local oppinion
given that the groups are connected by aquantinces that there would be
local pockets of variation of oppinion. However what I was getting at,
was could go through a real usecase that takes into account all the
varrious possible roles in this scheme?
Hmm, I'm not sure exactly what you mean... you suggesting a detailed
hypothetical?
Yes if we are going to seriously consider this, we should try to
describe it as accurately as possible:
A posts content offensive to B under SSK "A's site"
A's peers are A1, A2, A3.
After Pre-mix A is known as Z and has connections with Z1, Z2, Z3.
B sees it *somehow* and does not like it.
B's peers are B1, B2, B3.
After Pre-mix B is known as Y and has connections with Y1, Y2, Y3.
"A's site" was inserted through "()" represents tunnel:
(A1, A3, L7) Z1, H5, R6, L4, S2, S1
"A's site" is known to be stored on:
S1
who has peers:
S2, S3, S4
B's request path was "()" represents tunnel:
(B2) Y1, H2, R4, S3, S1
Given that:
Some nodes are offended by everything.
Some nodes are offended by nothing.
Some nodes never bother to check.
Some nodes are unattended.
Can you give the ensuing operations on the network, describing each hop
and what is involved? Then can we try to make a formulas for: bandwidth
consumed, hops required, number of people viewing A's site, how many
replicas of the content are made in this process, the amount of storage
required, the upper bound of using this as a DNS mechanism, the
effectiveness of using this as a "goatse troll" style attack on the
network, the upper bound on the effectiveness of using this with
colluding cancer nodes to find the identity of: A, B, and S, and when it
is all said and done the probability that A's site will actually be
brought down?
_______________________________________________
chat mailing list
chat@freenetproject.org
Archived: http://news.gmane.org/gmane.network.freenet.general
Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/chat
Or mailto:[EMAIL PROTECTED]
