1997-08-06 - Re: Eternity Uncensorable?

Header Data

From: Wei Dai <weidai@eskimo.com>
To: Mark Grant <mark@unicorn.com>
Message Hash: 7791154419549b952692cace0362bf3cfc06d8adc058f4a11a1fab6927283c08
Message ID: <Pine.SUN.3.96.970806112402.19588B-100000@eskimo.com>
Reply To: <Pine.LNX.3.91.970806145040.16936A-100000@cowboy.dev.madge.com>
UTC Datetime: 1997-08-06 19:03:55 UTC
Raw Date: Thu, 7 Aug 1997 03:03:55 +0800

Raw message

From: Wei Dai <weidai@eskimo.com>
Date: Thu, 7 Aug 1997 03:03:55 +0800
To: Mark Grant <mark@unicorn.com>
Subject: Re: Eternity Uncensorable?
In-Reply-To: <Pine.LNX.3.91.970806145040.16936A-100000@cowboy.dev.madge.com>
Message-ID: <Pine.SUN.3.96.970806112402.19588B-100000@eskimo.com>
MIME-Version: 1.0
Content-Type: text/plain

On Wed, 6 Aug 1997, Mark Grant wrote:

> That's not what I meant; I'm assuming that the mixmaster chain is pretty
> secure. The problem is that when the government find 'nuclear terrorist
> money-laundering kiddie drug porn' on the Eternity server they can trace
> it back to the original Usenet message and then go after the remailer
> operator who posted it. Making the remailers disposable or setting them up
> in free countries would work, but I'd prefer a technical solution. 

I suggest using information dispersal to spread risk amongst remailer
operators.  Use Rabin's information dispersal technique to divide up a
document into n shares such that k of them can reconstruct the original,
and post each share via a seperate remailer.  It would be hard for the
government to single out an operator to go after since an individual share
by itself is useless.

If n>k this also increases reliability and resilience of the eternity
service against technical attacks.