1995-02-10 - Re: Selection key crypto protocol trial balloon

Header Data

From: rishab@dxm.ernet.in
To: cypherpunks@toad.com
Message Hash: daca2459c57176947184238362c7983aed4c9573ccbf303e1bc5c9d6dcb29be0
Message ID: <gate.as70Zc1w165w@dxm.ernet.in>
Reply To: N/A
UTC Datetime: 1995-02-10 07:12:39 UTC
Raw Date: Thu, 9 Feb 95 23:12:39 PST

Raw message

From: rishab@dxm.ernet.in
Date: Thu, 9 Feb 95 23:12:39 PST
To: cypherpunks@toad.com
Subject: Re: Selection key crypto protocol trial balloon
Message-ID: <gate.as70Zc1w165w@dxm.ernet.in>
MIME-Version: 1.0
Content-Type: text/plain


Hal <hfinney@shell.portal.com>:
> Let me get straight where we are.  Rishab's concrete proposal was not an
> implementation, but rather a set of requirements.  There was no
> suggestion about any specific algorithms that would meet those
> requirements, right?  The question is whether any such algorithm could
> exist.

Right.

> It is hard for me to see how this could possibly work.  The message
> receiver sends this "selection key" to the intermediary, and that somehow
> pulls out the saved message, but in a form such that the intermediary
> doesn't recognize it.  And the intermediary himself can't tell exactly
> which message is produced.  But it is nevertheless exactly the message
> which was meant for this particular receiver.

It need not be the same information - i.e. if the encryption process adds 
enough noise to the message, the extraction process might by lossy, without 
any total loss for the receiver. This may be possible particularly if 
Fourier-style transforms are used, as the "selection key" could pick up 
less "frequencies" of a message than are stored at the intermediary.

> The thing is, the receiver does not have much more information than the
> intermediary.  

The same could be said of a private/public key pair. What is crucial is the
association between the keys.

This said, while I can imagine separate systems for the intermediary loosing 
correlation between incoming and outgoing messages (eg - it's _really_ tough,
without fancy pattern recognition, to associate a JPEG with the original), 
and for generating three associated keys (secret splitting? something to do
with Brands' blind signatures?), I really can't imagine a way of putting it
all together. Someone will have to be inspired, as Diffie was...


-----------------------------------------------------------------------------
For Electric Dreams subscriptions and back issues, send a mail to
rishab@arbornet.org with 'get help' as the message Subject.

Rishab Aiyer Ghosh          rishab@dxm.ernet.in           rishab@arbornet.org
Vox +91 11 6853410 Voxmail 3760335       H 34C Saket, New Delhi 110017, INDIA  





Thread