🚀 go-pugleaf

In <1992Sep28.095546.1@eagle.wesleyan.edu>, aspringman@eagle.wesleyan.edu
(Andrew Springman) writes:

> In article <1992Sep23.182419.21656@cactus.org>, ritter@cactus.org (Terry 
> Ritter) writes:
>
> > The way I understand it, Ross proposes to hide a small amount of
> > data in normal text by hashing 40-byte blocks into a single bit
> > using a secret key.  During text construction, the text editor
> > would "beep" if a just-created block did not produce the correct
> > bit; in that case the user would back up and try again with
> > different words.
>
>	You are right.  It does not solve the problem.  Every move of the
> person trying to distribute their key can be intercepted by the opponent in
> this model.  If the opponent can't tell what's up with a message, if the 
> outer message looks innocuous, he can change the wording of the message thus
> destroying the inner bit pattern without alerting the readers to a change.  
> The opponent could then, as you said, announce that the public key is in the 
> new message and publish a key and algorithm to find it (after the distributor
> attempts to announce such a key)

I believe I did mention in my original posting that the spooks did this 
during the war. If they saw a telegram saying `father is deceased' they'd 
change it to `father is dead'; if they got lucky, a dumb spy would query, 
`is father dead or deceased?' (This is all in Kahn's book.)

If the opponent isolates you totally, intercepts and modifies _all_ your 
communications, then you've indeed got a problem. But using steganography 
in the way I suggested can make his job much harder, especially if you 
couple it with a non electronic communications medium.

For example, if you write occasional articles, then you can hide your 
certificate in text which you email to the editor. Then stop to buy a copy 
of the magazine at a randomly chosen town; this should reassure all but the 
clinically paranoid that Big Brother didn't monkey with that particular 
email message.

I don't claim that steganography is a panacea, for the key certification 
problem or anything else. It's just one of the tools in the kit. If you're
going to use crypto to transfer trust, you may want to take a lot of care
about how you establish the first link in the chain of trust.

At the moment the military and diplomatic people do this with couriers, and 
the banks do it by writing letters to their correspondent banks' senior 
managers' home addresses; both can be subverted with enough effort. Using a 
broadcast medium like the internet, perhaps together with steganography, is 
one possible improvement; another might be for one trusted party, such as 
the Bank of England, to broadcast its key on teletext and the BBC World 
Service. You pays your money and you takes your choice.

However if the real goal is to shield oneself from harm at the hands of the 
State - whether deliberate or accidental - then cryptology is probably not 
the right approach. In that case, why not form a pressure group and raise 
twenty thousand a year for the local ruling party? That way you get access 
to ministers if you ever need it.

Ross