Re: pgp_anonvelope non-portability (was Re: pgpenvelope slowness) (fwd)

[Len S. <ra...@qu...>]

On Tue, 25 Sep 2001, Frank Tobin wrote:

> >From what I've read in order to get mixmaster support into mutt there are
> changes to the mutt source.  Not exactly a nice way to go about things,
> IMO.

Right, Mutt supports it natively, as it supports GnuPG and PGP. A far
nicer way to do things than filters IMO. But that's an old issue with Pine
vs. Mutt.

I can't see any reason pine would need to be modified for mixmaster
support. You could do it in a filter.

>    http://www.obscura.com/~loki/remailer/remailer-essay.html is a good
>    read.
>
> The attacks described against cypherpunk remailers have good defenses
> against them.  As long as the remailers are throwing around whitenoise all
> the time, the attacks can be heavily minimized.

The remailers throw around *type II* dummy traffic.

> The author also severely glosses over mixmaster in terms of security
> analysis.

You know he's the original author of mixmaster, right?

Another good place to read up on this is the cypherpunk mailing list
archives from around that time.

> Also, the attacks against cypherpunk remailers are primarily applicable
> against remailer-to-remailer communications, not MUA-to-remailer.  I think
> that a two-stage process is a good one, cpunk for the MUA-to-remailer for
> the first stage, and whatever the cpunk wants to do to get the message to
> the next hop is its own business; it could use mixmaster if it wants.

No, it doesn't work that way. The user has to create the chain and nested
encryption before sending the message. The remailers don't have control
over the format in which the user creates his message.