[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: exit node only server



On Fri, Mar 04, 2005 at 05:49:37PM -0500, Geoffrey Goodell wrote:
> You provide a rant, I provide a screed:
> 
> > At the risk of sounding like a broken record, I second (or third?)
> > this idea.  There are a number of impelling reasons for changing the
> > Tor architecture to a tit-for-tat construction:
> 
> You are confused.

Everybody is confused.  If you *aren't* confused about how to create
incentive structures for volunteer-operated anonymity networks, you
should let us know what the answer is.  :)

As noted by Chris, we are indeed hoping to more Tor to a more
decentralized discovery model.  We have to, otherwise user demand will
outstrip network capacity.  Chris is correct that a Bittorent-style
centralized t-f-t tracker would be problematical; it would present a
single point of failure, and create fairly bad anonymity problems by
potentially providing an adversary with a centralized view of how many
people have delivered how many bytes.  A localized mechanism may
have more hope, at some loss of efficiency--though it could still have
anonymity problems.

Just because something isn't in the codebase today, you shouldn't
assume that we have examined it carefully and determined it to be
unnecessary or ineffective---more likely, we have been very busy with
other things to keep the network up, running, happy, and as efficient
and featureful as we can make it.  (Though there *is* some stuff that
isn't there because we don't want it, so don't assume in the other
direction either.)

 [...Skipping discussion of human nature, t-f-t, routing versus
   swarming downloads, etc...]
> > In closing, I think it highly unlikely that Tor will ever be practical
> > on a scale much larger than the current experiment without changing to
> > a tit-for-tat design.  Tor networks will ALWAYS be resource-poor
> > unless tit-for-tat is at least semi-automatic (clearly, the
> > architecture should allow for server-only nodes).
>
> I dare you to say this in an objective manner.

Well, you could look at section 5 of our "Challenges" paper at
http://tor.freehaven.net/cvs/doc/design-paper/challenges.pdf .  It's
not been published yet, and there's as much speculation as anything
else, but hey, it could be a start.

Of course, it suffers from the same weakness as all of the other
discussions on this mailing list in that it doesn't actually propose a
design.  It sketches out the details of the outline of the shadow of a
design, however, which is better than no design at all.

The great thing about designs is that it's way easier to talk sensibly
about "Will this proposal work, and can it be fixed?" than it is to
talk about "Will any proposal containing this feature work?".

But yeah--don't worry if you think we might be idiots who aren't
thinking about incentives at all.  We are indeed thinking about
incentives.  We know they're important.  Honest.

yrs,
-- 
Nick Mathewson

Attachment: pgpUgOKxshvxN.pgp
Description: PGP signature