NODE ed656b95one time pads.
hkhenson@cup.portal.comMon, 19 Oct 92 02:21:17 PDT
I can suggest a way to "distribute" a one time pad, even though the
people never meet. Just agree over the phone on which CD ROM to use,
and some forumula for an offset into the CD ROM. You might want to
throw away some of the data to make the bit stream less regular, but
with 600 meg, who cares? Keith Henson
NODE 01dd3338one time pads.
Eric Hughes <hughes@soda.berkeley.edu>Mon, 19 Oct 92 08:48:48 PDT
Keith's CD-for-a-pad idea is a variant of a book code. In a book
code, parts of the key are in various standard books, often the bible.
Advantages: easier key distribution.
Disadvantages: key material is public. Should an internal spy learn
the few bits of addressing information (which CD, where), the cipher
is compromised.
Eric
NODE aa4ce897one time pads.
pmetzger@shearson.com (Perry E. Metzger)Mon, 19 Oct 92 17:09:14 PDT
>From: Eric Hughes <hughes@soda.berkeley.edu>
>Keith's CD-for-a-pad idea is a variant of a book code. In a book
>code, parts of the key are in various standard books, often the bible.
>Advantages: easier key distribution.
>Disadvantages: key material is public. Should an internal spy learn
>the few bits of addressing information (which CD, where), the cipher
>is compromised.
Actually, in practice internal spies were almost never needed to break
book cyphers. They in fact provide only laughable security.
Traditional ones didn't even require that the cryptanalyst ever
determine what book was being used!
Perry
NODE 982bfab0one time pads.
pmetzger@shearson.com (Perry E. Metzger)Mon, 19 Oct 92 17:08:26 PDT
>From: hkhenson@cup.portal.com
>I can suggest a way to "distribute" a one time pad, even though the
>people never meet. Just agree over the phone on which CD ROM to use,
>and some forumula for an offset into the CD ROM. You might want to
>throw away some of the data to make the bit stream less regular, but
>with 600 meg, who cares? Keith Henson
This seems equivalent to the old "dictionary" or "book" cyphers that
people sometimes used. Good cryptanalysts broke them routinely. I'll
leave it to your imagination how one might do it, but I'll just note
that if you picked a few arbitrary bytes, say bytes 30-40, of all the
CDs in the record store, you would find that those few bytes likely
distinguish all but prehaps a token number of CDs.
Perry