Saturday, January 07, 2012

Gouzenko and the Soviet target

The defection of Igor Gouzenko on September 5th, 1945 was one of the key events in the outbreak of the Cold War. Gouzenko, who was a GRU (Soviet military intelligence) cipher clerk at the Soviet embassy in Ottawa, brought with him evidence of massive Soviet spying in Canada, including penetration into the Manhattan Project, the U.S. effort to build the atomic bomb. (More on Gouzenko here.)

Gouzenko’s revelations, which were quickly shared with the U.S. and British governments, sent shock waves through all three capitals.

But not because any of these governments found the idea of allies spying on allies inconceivable. The U.S. had been intercepting Soviet communications traffic and attempting to break into Soviet diplomatic codes since 1943. And the United States and Britain had agreed on August 15th, 1945, three weeks before Gouzenko’s defection, to extend their highly successful wartime SIGINT partnership into the post-war era, with the target of their cooperation being Soviet communications.

Thus, when Gouzenko defected, his knowledge of Soviet cryptographic procedures and techniques was also of great interest to the Western allies.

The extent to which that knowledge was of assistance to Western SIGINT efforts, including the now-famous VENONA code break into KGB, GRU, and Soviet diplomatic communications, remains largely hidden, but there has been some discussion of Gouzenko’s contribution in declassified documents.

The VENONA effort began in the U.S. in 1943, but it took a number of years before texts could be even partially read, and even in 1980, when the project was closed down, many gaps still remained.

An unclassified official history of the VENONA project (Robert L. Benson, The VENONA Story, National Security Agency: Center for Cryptologic History, 2001) reports that Gouzenko made no direct contribution to the project:
While the 1945 defection in Ottawa of GRU code clerk Igor Gouzenko provided message texts that revealed a great deal of espionage, Gouzenko did not produce any cryptographic materials of direct use to Arlington Hall’s cryptanalytic effort. In the long run, success against GRU messages came from the accumulated knowledge and experience gained exploiting the Soviet trade, true diplomatic, and KGB systems and the application of early computers. (p. 44.)
However, NSA’s top-secret history of Cold War cryptology, released to the public in redacted form, is a little less unequivocal (Thomas R. Johnson, American Cryptology during the Cold War, 1945-1989: Book I: The Struggle for Centralization, 1945-1960, National Security Agency: Center for Cryptologic History, 1995, Top Secret Umbra, released in redacted form):
...after World War II several outside factors speeded the tortuously slow process of additive key diagnosis and recovery and bookbreaking. The first was the defection of a Soviet GRU cipher clerk, Igor Gouzenko, from the Soviet Embassy in Ottawa, in September 1945.... Because Gouzenko worked with communications, Frank Rowlett of ASA was invited to interrogate him. During his sessions Rowlett learned much about the way the KGB codebooks were put together and how the additives were used. This information cut time off ASA’s cryptanalysis effort. (p. 161)
Robert Louis Benson and Cecil J. Phillips’ History of Venona (National Security Agency: Center for Cryptologic History, 1995, Top Secret Umbra, released in redacted form) seems to have been Johnson’s primary source on Gouzenko. The History of Venona provides considerable detail about the information provided by Gouzenko and is worth citing at length:
Lieutenant Colonel Frank B. Rowlett departed Washington, in civilian clothes, on 25 September 1945, to question Lt. Igor Gouzenko, the GRU code clerk who had defected....

Frank Rowlett spent several days questioning Gouzenko, codenamed “Corby”, and the following paragraphs are based on his “Special Report on Bourbon Cryptography: Report on Interrogation of Corby”, dated 15 October 1945. [Bourbon was the codeword for the U.S.-British effort against Soviet communications at that time.]

...

Rowlett drove to an isolated, lakeside summer cabin some 90 miles from Ottawa where Gouzenko was under guard. The Rowlett party included Professor Gilbert Robinson, a wartime Canadian Sigint officer, Inspector Leopold of the RCMP and a driver. (Robinson had conducted the preliminary questioning of Gouzenko on cryptologic matters – Gouzenko had given names of spies and supporting papers to the Secret Service [sic; U.K. Security Service?], RCMP and FBI.)

Rowlett learned that Russian cryptography, in the external affairs area, could be divided into two types:
  • Systems used by Russian establishments abroad in communication with Moscow. These systems were entirely by encipherment of a code by a one-time additive.

  • Emergency or illicit systems which used a substitution alphabet based on one and two-digit equivalents for the Cyrillic alphabet, which would then be enciphered by one-time key generated from a book or other publication readily accessible to both Moscow and the communicant in the field, i.e., both Moscow and the field had to have the same book, same edition.
Gouzenko explained Russian crypto-security doctrine and procedures, and the day to day work of a GRU code clerk. All code clerks were approved and trained by the KGB. Every Russian official authorized to sign messages—the GRU Resident/Military or Naval Attache, the Trade Representative, the Ambassador or Consul—had a code clerk assigned to him who would prepare the messages. These clerks were responsible to the KGB for security and procedures—the officials who drafted the messages could not keep file copies of the original texts of the messages they were releasing. In the case of the GRU Resident in Ottawa, he would bring his notes into the office of the code clerk, and in the presence of the code clerk write out a message. The code clerk, after the drafter had left the office, would encode the message and then take it to the mission’s communications officer who would give it an external serial number and take it to the commercial cable company for transmission to Moscow (the Russian establishment in Ottawa did not at that time have its own communications facilities).

The Russian text of a GRU message would be encoded by a four-digit, one-part code, that is, the codebook was arranged in strict alphabetic order. When an item had to be put in the message for which no equivalent appeared in the code book, it was spelled out by means of a Cyrillic or Roman substitution alphabet which was issued as a supplementary chart to the code. When this chart was to be used the four-digit group 7810 would be entered, meaning “begin spell” and the end of the spelling would use a special two-digit group, 91, to mean end of the spelling.

The GRU code text would then be enciphered by a one-time pad. All pads, for every agency of the Soviet government, were manufactured by the KGB. The pads had either 35 or 50 pages each and each page would contain ten lines of five 5-digit groups, for a total of 50 groups or 250 digits per page of pad. Each page or sheet of the pad had a two digit number in the upper left hand corner ranging from 01 to 35 or 50 depending on the number of pages in the pad. These pads were carefully packaged and controlled. When the message reached Moscow, a senior officer would determine the addressee and pass it to the appropriate code clerk. Gouzenko reported, curiously, that copies of these GRU messages would go directly to the KGB for analysis (perhaps he meant, also to the KGB). Gouzenko described for Rowlett, at some length, the emergency or illicit systems.

Gouzenko believed that the KGB, in producing all one time pads for the government, mechanically generated them using an apparatus, “which selects numbers purportedly at random by a device using small balls in some fashion. This apparatus was credited to the British originally, but certain improvements were made by the (KGB) cryptographers when it was adopted by them. No further information regarding pad generation is available.”

...

A small team in Bill Smith’s Russian unit consisting of Mrs. Genevieve Feinstein, Miss Mary Jo Dunning and Mr. Burton Phillips immediately began a study of the Rowlett report in context of the traffic on hand. It is maddening for the non-cryptanalyst to try to understand, but then to realize, that even with such a source as Gouzenko, who brought out plaintext of some of the GRU encrypted messages and explained the system in great and accurate detail, the traffic remained unbreakable. Gouzenko’s background information on the Russian systems was certainly very important and helpful—but with it Arlington Hall could not read any traffic and could at best only add some words to book breaking vocabulary of the GRU code book. But the code book would not do anything unless the cipher additive, from the one time pads, could be identified and stripped off to reveal the underlying code groups. Gouzenko had no pads, and if he had it would only have given an opening into the message(s) enciphered by that particular pad. Gouzenko’s most enduring contribution to Venona was to put the cryptanalyst into the office of a Russian code clerk, giving us an understanding of how he worked, and what his systems looked like and how they were used. (pp. 61-64)
It is likely that Gouzenko’s information on Soviet cryptologic techniques and doctrine was useful to more than just the VENONA effort. His information on emergency/illicit communications procedures, for example, was probably of great use in counter-intelligence investigations and may have contributed to the decipherment of some messages.

Based on these reports, it does not appear that his information made much direct contribution to the VENONA project. His contribution to the book-breaking element of the project was, as noted in the official histories, probably of some value, however. Gouzenko did not bring a GRU code book with him when he defected, but he would certainly have known the code groups for many of the most commonly used words in the book. This information, and the information that the GRU book was a one-part (alphabetical) code, would probably have been of considerable help to the book-breakers when they began the job of recovering the meanings of the code groups in the GRU code book.

It seems possible (at least to me) that Gouzenko’s information on the Soviet spell/endspell procedure was also helpful. Meredith Gardner’s exploitation of spell/endspell sequences led in December 1946 to one of the key early breaks in the VENONA project when he uncovered the names of Manhattan Project scientists in a KGB message. Presumably most, if not all, such codes have some sort of spell/endspell capability, so Gardner would almost certainly have made the break regardless. But it may be that Gouzenko’s information eased the process somewhat.

[Update 1 February 2012:

According to John Bryden's book Best-Kept Secret: Canadian Secret Intelligence in the Second World War (Lester Publishing, 1993, p. 277) "Mackenzie King did report in his diary that Gouzenko was believed to have taken a code book." If the NSA histories are correct, however, Gouzenko did not bring a code book with him.]

0 Comments:

Post a Comment

<< Home