From: Adam Back <aba@dcs.ex.ac.uk>
To: paul@fatmans.demon.co.uk
Message Hash: cdf81b323321e832ed18ea4a9a0604bf6b4798e7f66f56437c610baf8e39a974
Message ID: <199705151813.TAA01081@server.test.net>
Reply To: <Pine.LNX.3.91.970513165201.835A-100000@fatmans.demon.co.uk>
UTC Datetime: 1997-05-15 18:26:20 UTC
Raw Date: Fri, 16 May 1997 02:26:20 +0800
From: Adam Back <aba@dcs.ex.ac.uk>
Date: Fri, 16 May 1997 02:26:20 +0800
To: paul@fatmans.demon.co.uk
Subject: Re: Eternity server considerations and musings
In-Reply-To: <Pine.LNX.3.91.970513165201.835A-100000@fatmans.demon.co.uk>
Message-ID: <199705151813.TAA01081@server.test.net>
MIME-Version: 1.0
Content-Type: text/plain
Paul Bradley <paul@fatmans.demon.co.uk> writes:
> The main method used by governments around the world so far to
> attempt to censor information has been the use, or proposal of,
> proxy servers for whole countries or jurisdictions. This method has
> only thus far been used to censor www connections, and has been used
> to censor specific sites, hence the use of mirroring to circumvent
> government censorship.
One of the main considerations in the design of my eternity server was
the idea that there was information which might get published via
remailers, but would be unlikely to get published on a web page
directly. Eternity translates the ability of publishing via a
remailer into the ability to publish on the web.
An example is perhaps the NSA handbook, or Mykotronix dumpster diving
results. These sorts of things tend to arrive via remailer, and then
get mirrored afterwards.
There will also be some class of documents which are too hot to mirror
at all, or where the censor is litigious enough (Scientologists?) to
present a real danger to it's ciritics, and to mirrorers of the
criticisms.
A question then is could a censor systematically block all eternity
search engines effectively. I think this would be relatively easy,
the design does not really attempt to deal with site blocking
directly.
The URL always includes "eternity" so is easy to block. You could
perhaps provide an option for a public key encrypted URL. Or lookup
by the SHA1 of the URL directly. An SSL session would also help, the
cgi-binary accepts both post and get methods. However ultimately, the
eternity servers will be advertised, and so the censors just need to
keep track and block all the advertised servers.
However there is another option: all the data is available in USENET
anyway, so anyone can run an eternity search engine for their own use
in a shell account. To block this is more difficult.
> [blocking USENET, key word searches, recognizing encrypted data]
Use stego solves this (as Bill Frantz observed). Do a web search on
`texto', it's a nice simple text stego program.
Another interesting fact is that they can't block the traffic unless
they have decrypted it. As the data can be encrypted with the SHA1 of
the URL, they won't necessarily recognize it as an eternity web page
until it gets accessed or the URL advertised, by which time it is too
late. The URL can be smuggled in and passed around more readily.
> [a censor can provide a censored USENET feed for their country, they
> can't provide a censored WEB feed because it's too large]
WEB mirroring is useful where the data being mirrored is illegal in
someone else's country and you are trying to stay one step ahead of
the censor.
Eternity means you can publish material which is illegal or
dangerous/unpopular to publish in your own country.
> [...] it would be nice to see a more robust system which thwarted
> all attempts to censor without the need for human intervention by
> mirroring sites etc.
Sounds like something that could be addressed by building on the
rotating mirror idea William Geiger has been talking about. Perhaps
you could have an apache module which provided a proxy service which
allows encrypted and steganographically encoded URL passed to it, and
encrypted, stego web pages passed back. Combine with some unstego
method on the receiving end, and lots of people using apache with this
option turned on, and the censor would have one heck of a problem on
their hands.
Adam
--
Have *you* exported RSA today? --> http://www.dcs.ex.ac.uk/~aba/rsa/
print pack"C*",split/\D+/,`echo "16iII*o\U@{$/=$z;[(pop,pop,unpack"H*",<>
)]}\EsMsKsN0[lN*1lK[d2%Sa2/d0<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<J]dsJxp"|dc`
Return to May 1997
Return to “Paul Bradley <paul@fatmans.demon.co.uk>”