[Cryptography] [Crypto-practicum] An historical document
agr at me.com
Mon Sep 12 14:36:57 EDT 2016
On Sun, 11 Sep 2016 12:20, Henry Baker wrote:
> You couldn't find more fertile soil than Turing's
> mind circa 1937; even the merest hint connecting
> tabulating machines and coding would have been all
> that was required for him to see all of the
> Turing would certainly have known about punched
> card tabulating machines; I believe that they
> were also in use in England (but with round
> holes?). He very likely saw them at Princeton
> or during his travels in the U.S.
> There were various schemes already in use for
> hand sorting edge-punched cards with long thin
> steel rods; Knuth covers these schemes in his
> Punched paper tapes had already been used for
> ~50 years for teletypes and stock market tickers.
> I worked on these so-called "Electronic Accounting
> Machines", which included *sorters*, *mergers* and
> Sorters were read-only devices that rearranged the
> ordering of the cards. Mergers (programmed with
> wired plug-boards) would merge 2 (or more) decks
> of sorted cards and merge the information from
> the two (or more) different streams into an
> newly punched output stream. Printers (also
> programmed with wired plug-boards) would process
> sorted card decks and subtotal certain columns on
> a "break", where a high order field in the sort
> order would change.
> I seem to recall that only the printers were
> smart enough to do arithmetic; the mergers could
> compare, but I don't recall their ability to
> add or subtract (at least in the earlier models).
> P.S. Wouldn't it be a scream to find out that
> if Turing *hadn't* seen this Coast Guard publicity,
> he might never have been interested in coding and
> Britain would have lost the war!
> In this particular case, secrecy might have killed
> the very technology that would have won the war.
> I'm afraid that crypto secrecy -- in general -- set
> everyone back far more than it helped. Look at how
> the field exploded once Diffie/Hellman/RSA got
Alan Turing's “Turing Machine” paper, "On Computable Numbers, With an Application to the Entscheidungsproblem," was delivered in November 1936, so he was already thinking deeply about the possibilities of machine computation. His biography on Wikipedia says "From September 1936 to July 1938, Turing spent most of his time studying under Church at Princeton University. In addition to his purely mathematical work, he studied cryptology and also built three of four stages of an electro-mechanical binary multiplier.” (citing Andrew Hodges’ biography of Turing)
Also the Polish cryptographer Marian Rejewski invented the Bomba in 1938 and told the British about it in 1939. Truing was working part time for the British GC&CS from September 1938 so even in the extremely unlikely event that he would not have thought of using machines to break codes, news of the Polish success would have gotten him going.
The IBM punched card equipment you talk about was being actively sold in England and Germany in the 1930s. What you call printers were sophisticated tabulating machines, which could be programed much like FPGAs today, i.e. they had a variety of relay logic components and electro-mechanical adder-counters that could be wired up to perform calculations. See the Wikipedia article on Plugboard, which describes how they work in some detail. As far as I know, the US cryptologic groups used them to tabulate statistics about intercepted ciphertext and codes, not for testing large numbers of possible key settings at high speed, which is what the Turing Bomb and Colossus did.
Finally IBM when I knew them was very carful about getting permission from customers before touting their use of IBM equipment in ads. I suspect they were as careful in the 1930s. I have to agree with bear that this very interesting memo is more about insuring that the use of tabulating equipment for cryptanalysis stays secret than a response to a leak that Turing would have picked up and by doing so changed history.
More information about the cryptography