gts@dmcnh.UUCP (11/16/84)
+[ The mailer deamon wouldn't eat this, so I'm posting it: ]+ I mentioned that Beale's cipher #3 had been determined to contain a message, but that it hasn't been cracked. I got this letter, but could not respond via mail: > Subject: Re: Chuq's comments about ESP > What are the statistical methods for determining the existence of > a message? (I have not studied cryptography very much) > [tectronix!sharkp!brian(?)] I must admit that I only know of the existence of such methods. For more information there is an excellent Journal you might try, called _Cryptologia_ (it comes out quarterly): Dr. Brian Winkel Editor Cryptologia Rose-Hulman Institute of Technology 5500 Wabash Ave. Terre Haute, IN 47803 Dr. Winkel, before becoming a mathematics professor at Rose-Hulman, worked for the National Security Agency (NSA) which is in possesion of more comp- uter power than any single installation in the free world. They sit all day long intercepting messages and trying to decode them (usually success- fully). In fact, just mentioning NSA in this letter will probably bring it to the attention of someone besides just you and me, if you catch my drift. Anyway, any specific questions about Cryptology could also be addressed to Winkle personally, although I'm not sure what result you may get. -From the padded terminal of ><..!decvax!ittvax!sii!dmcnh!gts Disclaimer: This posting is the sole responsibility of myself and does not reflect the policies of Datamedia Corporation.
dee@cca.UUCP (Donald Eastlake) (11/21/84)
For any particular set of statistical measures, you could obviously
generate a random message that met them and would thus seem to contain a
message, even though it didn't.
--
+ Donald E. Eastlake, III
ARPA: dee@CCA-UNIX usenet: {decvax,linus}!cca!deegwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (11/21/84)
> For any particular set of statistical measures, you could obviously > generate a random message that met them and would thus seem to contain a > message, even though it didn't. I seem to recall an article about random message generation using given uniliteral, digraph, trigraph, etc. distributions not long ago in the "Computer Recreations" column in Scientific American.
jlg@lanl.ARPA (11/21/84)
> For any particular set of statistical measures, you could obviously > generate a random message that met them and would thus seem to contain a > message, even though it didn't. > For any particular set of statistical measures, you could obviously generate a real message with a pseudo-random encryption scheme which would seem not to contain a message, enen though it did. The problem here is that you must know the set of statistical measures first (or be VERY lucky).
keithd@cadovax.UUCP (Keith Doyle) (11/21/84)
>For any particular set of statistical measures, you could obviously >generate a random message that met them and would thus seem to contain a >message, even though it didn't. > + Donald E. Eastlake, III Sounds like a good way to keep someone busy for a long time.
herbie@watdcsu.UUCP (Herb Chong, Computing Services) (11/22/84)
I may be a bit shaky on this point, but aren't you confusing message
(information transfer) and meaning (semantics)? The "randomly
generated message" carries a great deal of information about the
message generation process but does not have any meaning associated
with it. This is like transmitting data in a foreign language which
can convey a great deal about the structure of the language but almost
none of the semantics.
Herb Chong...
I'm user-friendly -- I don't byte, I nybble....
UUCP: {decvax|utzoo|ihnp4|allegra|clyde}!watmath!watdcsu!herbie
CSNET: herbie%watdcsu@waterloo.csnet
ARPA: herbie%watdcsu%waterloo.csnet@csnet-relay.arpa
NETNORTH, BITNET: herbie@watdcs, herbie@watdcsu
POST: Department of Computing Services
University of Waterloo
Waterloo, ON
N2L 3G1 (519)885-1211 x3524