[14310] in Cypherpunks
Re: (fwd) Re: NSA Helped Yeltsin Foil 1991 Coup
daemon@ATHENA.MIT.EDU (David Miller)
Fri May 27 16:17:03 1994
Date: Fri, 27 May 1994 16:11:01 -0400 (EDT)
From: David Miller <isdmill@gatekeeper.ddp.state.me.us>
To: Eli Brandt <ebrandt@jarthur.cs.hmc.edu>
Cc: cypherpunks list <cypherpunks@toad.com>
In-Reply-To: <9405271955.AA21973@toad.com>
On Fri, 27 May 1994, Eli Brandt wrote:
> > From: "Perry E. Metzger" <perry@imsi.com>
> > > If you suspect that some of the non DOD/NSA cyphers might be broken,
> > > but you are not ready to employ one-time-pads, then you should
> > > threshold you mesages into N parts so that all N are needed to recover
> > > the original. Then encrypt each part under a different cypher.
> >
> > Its far simpler to encrypt your message with multiple systems, one
> > after another, than to break it up in the manner you suggest, and the
> > security is in fact better that way than in the manner you suggest.
>
> Why? If you XOR-split the message and encrypt each mask differently,
> you are /guaranteed/ that all of the encryption methods must be
> broken to retrieve the original. If you use repeated encryption,
> this is much harder to prove, and not always true. There's a result
> that if you choose the first cipher unwisely, you're hosed no matter
> what you do on top of it.
>
> Eli ebrandt@hmc.edu
I think the second poster assumed what I did - that the message would be
split into say 5 parts, each to be encrypted differently. How to X-or
split the message isn't obvious to me - pnrg? If you use some bytes
conveniently hanging around you may as well use a OTP, since both ends
need the same bitstream. Unless I'm missing something, which is usually
the case.
David isdmill@gatekeeper.ddp.state.me.us
