Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:15:20
Jim McCoy (mccoy@yahoo-inc.com)
Wed, 7 Oct 1998 14:31:47 -0700
Dan Bailey wrote:
>Pointers appreciated to a working searchable archive of the
>Cypherpunks list from '94-'95 or so. I'm interested in articles relating
>to DC-net iplementation, especially use of reputation
>capital to punish disruptors, which I (seem to) recall seeing there long
>ago. I've already got all the formally-published articles
>INSPEC can find, now I'm looking for some of the old list
>discussion.
Using reputations to regulate DC-nets wouldn't be there because the idea of
dealing with disruption in this fashion was first made public around mid '96
to early '97 (being the person who came up with it I think I should know :)
and other than a single posting to CodherPlunks a while back when DC nets were
being talked about, a section in a talk I gave at FC97, and a write-up of a
conversation over lunch between myself and Simon Spero I do not think that
there is much online about the system.
The idea is really quite simple: disruption is noise, there are very good
systems that lots of smart people have worked on to deal with noise within a
network (think ethernet and IP); treat disruption on the system as noise and
use a reputation market for deciding who will be allowed to join DC-net
cells. Think of it as a sort of bandwidth futures market, you negotiate
with other participants to create a cell by declaring how much bandwidth you
will pass. If everyone is pretty close in how much bandwidth they can pass
the cell is formed and the system throughput is compared to the theoretical
throughput. If the two numbers differ by too much or a pre-determined timer
expires the cell disbands and everyone marks their internal "reputation
scorecard" for the cell participants. They then form a new cell and use the
reputations of other participants to determine who can join the cell.
Given that the system provides for the security and anonymity of the
participants a disrupter who allows even a small amount of packets through
is still performing a small amount of service to the other members of the
cell (anyone who simply refuses to pass packets or disrupts all
transmissions will cause the cell to collapse instantly.) All cell
participants are scored the same, but because the mix of cell participants
should be random each time a cell is formed a participant will build up a
reputation over time regarding how trustworthy they are and/or how much
bandwidth they can reliably deliver.
The downside is that it requires a fairly large pool of participants, an
external channel for negotiating cell formation and breakup, and some sort
of barrier to entry to prevent disrupters from using disposable identities.
These are all things which are relatively easy to deal with and the system
avoids all of the nasty (and incredibly expensive) disruption detection
mechanisms which make up most of the work done on DC-nets after Chaum's
original paper.
jim
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:15:20