From: Matt Blaze <mab@research.att.com>
Date: Wed, 8 Feb 95 15:35:55 PST
To: weidai@eskimo.com
Subject: Re: a new way to do anonymity
In-Reply-To: <199502082243.AA19005@mail.eskimo.com>
Message-ID: <9502082318.AA21943@merckx.info.att.com>
MIME-Version: 1.0
Content-Type: text/plain



...
>since you can use it to do nested encryption.  All that's needed is 
>to hack it so that it implements link encryption (i.e., send a
>constant stream of random data in between keypresses).
...
You could just send a stream of some uncomon ascii character, which you
filter out on the receiving end (if you wanted to this right, you could add
a simple escape mechanism for actually passing that character).

To avoid flooding the network and also bringing the machines on which its
running to its knees, you'd probably want to add a bandwidth-choke
mechanism to run the white noise at some reasonable rate.  You'd have to limit
the real traffic output to the same rate.  Link encryption over a broadcast
network is a tricky business.
>
>I wonder if Matt has the time and interest do this...  If not then I 
>guess I can try, but I've never done real crypto programming before...
>

For the next couple of months, I have absolutely no free hacking time.
Things on the stack include:
	- ESM 1.0
	- Diffie-Hellman encrypting and authenticating Telnet (almost ready...)
	- CFS 1.3
	- The course
	- The book
	- My real work

So I don't even have the time to figure out whether I have the interest.

-matt