[Cryptography] RIP Claude Shannon

Jerry Leichter leichter at lrw.com
Tue Feb 23 17:30:41 EST 2016


>> Claude Shannon was lost to us on this day in 2001; he is regarded as the father of information theory, viz: given a noisy comms channel then there are this many bits that you can squeeze through it at most.
>> 
>> He also did top-secret crypto work in WW2.
> 
> Supposedly, Shannon worked on voice encryption for phones (a loser, in light of digitization & digital encryption)
Isn't it great to be able to sit here comfortably in the future, looking back at how silly our predecessors were?  Why bother with voice encryption for secure communications during WW II when all you have to do is wait, oh, 15-20 years for the concept to be developed, and 40 or so years for it to be really practical?

BTW, until Shannon, it would have been difficult for anyone to conceive of digital voice transmission.  The idea of "information" as a generic thing that could be transmitted in many forms wasn't there yet.  There were completely independent technologies to deal with voice and with Morse code.  (I'm not sure what other modes of electronic communication existed at the time.  Certainly there were remote sensors and actuators - selsyns's and such - but they would probably not have been viewed as being involved in "communication".  It was Shannon who introduced the common notion of "communication" and showed that the fundamental issues were independent of domain.  His breakthrough was similar to the emergence of a unifying notion of "energy" from separate ideas of energy of motion and potential energy and heat and electromagnetic energy and so on.

> and almost certainly on attempting to decrypt 1x pads -- the precursor to the Venona project.
> 
> It was probably Shannon's job to deliver the message to the top brass that the Soviet 1x pads were indeed unbreakable, and -- short of finding a copy of the pad or evidence of pad reuse -- that 1x pad-encrypted messages would leave the U.S. in the "dark".  I can imagine the top brass then going into a hissy-fit and asking all of those PhD's to work *harder* on this problem -- a la James Comey & Hillary Clinton.
The unbreakability of a 1-time pad is a pretty obvious notion and easy to understand.  Shannon's proof is important more for his way of formalizing the concepts to the point where a mathematical proof could be written down than in what it actually proved - I doubt any real crypies needed convincing.

Book codes had been known for many years, and methods for attacking them had also been known - which meant it was understood that to have an attack, you needed a handle on the statistics of the "book".  No statistics, no attack; and even then, you needed repeated messages encrypted against the same book to make practical progress.  What was needed to make Venona - where the "book" had flat statistics - attackable, and the general form of an attack, would have been well understood.  The details of making it practical ... that's another story, and one that Shannon may well have been involved in.  (I don't know if anything about this has been declassified and published.)

At the time Shannon was working, we were fighting for our lives.  That tends to bring out actual leaders and actual understanding of the realities.  While politics is never absent, it's usually submerged somewhat when the survival of the country is at stake.  (At least it is in countries that survive....)

                                                        -- Jerry





More information about the cryptography mailing list