Re: [freehaven-dev] re: chapters done -- moving forward options

[Roger Dingledine <arma@mit.edu>]

On Thu, Dec 14, 2000 at 07:03:00PM -0500, Michael T. Shinn wrote:
> > b) Implementing the current Free Haven design. I recall a number of
> > people were interested in this earlier, but they very conveniently
> > disappeared after voicing their interest.
> 
> Is this described in the papers on the free haven website?
  
Yes. You can get a pretty good description from the current draft
of the Berkeley paper.

It so happens that the final copy for the Proceedings is due on
Saturday. So that's what I'm up to currently.

Basically I will be combining
http://freehaven.net/doc/berk/freehaven-berk.ps (which
is technically detailed but slightly out of date) with
http://freehaven.net/doc/oreilly/freehaven_2.ps (which is newest but
glosses over some things).

> > c) Integrating strong anonymity into Freenet or Mojo Nation, and making
> > sure they get the design right.
> 
> This might seem unimportant to the process of design, (and someone else
> might have already said it) but I would suggest integrating strong
> anonymity into Freenet simply because of the media attention it has
> recieved.  Perhaps I'm assuming too much, but I would suspect the intent is
> to get as many people as possible to use free haven.
 
This is trickier than it sounds. First of all, we don't really want people
to start using Free Haven right now, because it isn't "good enough". We
think it's better than anything else out there -- at what it does -- but
it still isn't what we want. Our main goal at this point (regarding Free
Haven) is to research as many of the problems as possible and start trying
to solve them in a way that's useful to everybody else out there too.

Secondly: the Freenet design is not conducive to layering on stronger
anonymity.  They're not 'bad' at present -- they provide something more
than Gnutella, eg -- but their approach is a very practical one which
is very tough to assess and analyze. It *might* work -- but against
who? Freenet will be fine as long as nobody really big tries to break
it. At that point, who knows.

The Mojo Nation design, on the other hand, has hooks for where we
might make transactions more anonymous. It's built in a more rigorous
framework -- it would be challenging to do, but it's clear where to
do it.  On the other hand, the response when I asked the Mojo Nation
leader about anonymity was that "customers don't want it, so it doesn't
matter." That's a shame. (Mike Freedman's response to that was "ok. if
they want to be market-driven and corporate, they can be market-driven
and corporate by themselves.")

Erm. Did I answer any questions, or just create new ones?
--Roger