Does anyone know anything about James Radford (James Radford - Jim Radford developed Special Purpose Devices that solved intractable analytic problems, often by enhancing the performance of supercomputers by a factor of hundreds)?
I found that super interesting but couldn't find anything online about him, not even the era (Cold War, 911, recent). If anyone has any links or literature about things he did I would very much appreciate it.
While that list contains many important contributors, it is far from complete.
The most notorious of the 2023 list is Joseph Mauborgne, and among his merits is written: "He is credited as the co-inventor of the One-Time Pad".
Even if this claim, which appears to originate from the book "The Codebreakers" by David Kahn (1967) has been frequently repeated, there is no evidence for it and it seems very implausible.
The "One-Time Pad" has been described for the first time in the non-classified literature in February 1926 by Gilbert Sandford Vernam in “Cipher Printing Telegraph Systems For Secret Wire and Radio Telegraphic Communications”. Because of that, it has become known as the "Vernam cipher", even if Vernam has not invented it.
The "One-Time Pad" is an improvement of the so-called "running-key ciphers", which had been used already for many years before World War I. These are aperiodic substitution ciphers. Until the end of WWI it was believed that if the "running key", i.e. the stream of key symbols, is not periodic, that is enough to make an unbreakable cipher. The "running key" used for encryption was usually taken from the text of some book.
In 1918, two employees of ATT, Gilbert Sandford Vernam and Lyman F. Morehouse have filed two patent applications for an electromechanical implementation of the running-key ciphers, where the plaintext, the ciphertext and the running key were stored on punched tape. Vernam's patent was for the use of bitwise addition modulo 2 for combining the running key with the plaintext or ciphertext, while Morehouse's patent was for using several running-key generators with coprime periods and combining their outputs to obtain a generator with a period equal to the product of the coprime periods.
Both inventions of Vernam and Morehouse continue to be used today very frequently and they both deserve to be included in NSA's Cryptologic Hall of Honor more than most people already present there.
In 1918, Vernam and Morehouse who were in contact with Mauborgne, because the US military was a very likely customer for their encrypted telegraph, were still believing that it is enough for the running key to be not periodic. Some time between 1918 and 1926, Vernam has learned that there is a second condition, the key symbols must be chosen at random, otherwise the cipher is breakable.
Kahn supposes that Vernam has learned this from Mauborgne. This is possible, but in any case the idea cannot have originated from Mauborgne, but only from his subordinate Captain William F. Friedman.
NSA's Cryptologic Hall of Honor includes, very appropriately, at its first two positions (i.e. in 1999), both William F. Friedman and his wife and coworker Elizebeth S. Friedman.
In 1918, at the end of WWI, William F. Friedman has been the first who has succeeded to cryptanalyze documents encrypted with aperiodic running-key ciphers, busting the myth that such ciphers are unbreakable. Being the first who has created a deciphering method for aperiodic substitution ciphers that was based on the fact that the key symbols were not random, it is pretty obvious that he was also the first to understand that an unbreakable cipher must satisfy 2 conditions: the stream of key symbols must be aperiodic and random.
His work was classified, so a few months later Vernam and Morehouse were still believing in the unbreakability of aperiodic running keys, regardless whether they are random or not.
Mauborgne was Friedman's boss, so he must have learned immediately that the non-random running-key ciphers are breakable and that random aperiodic running-keys are required for unbreakability.
In the following years Friedman has collaborated with Vernam and he has invented some improvements of Vernam's system, to make its running keys more random.
So Vernam could have learned about the randomness condition either directly from Friedman or through Mauborgne. In any case, it seems impossible for Mauborgne to have had any direct contribution to the previous work of Friedman.
Even if it is unlikely that Mauborgne has been any kind of "co-inventor of the One-Time Pad", he certainly had very important contributions so he deserves his place in the Hall of Honor. Nevertheless, nothing of what Mauborgne might have done is still in use today, unlike the inventions of Vernam and Morehouse, which are ubiquitous, so they deserve more than him a place there.
Also Shannon (ATT), Hamming (ATT) and Feistel (IBM) are missing, while all modern cryptology is based on their work (Diffie is included on the list, despite his constant mistrust of NSA, so the list is not restricted to government employees). It is less known that Hamming had an essential contribution to modern cryptology. While his colleague Shannon had invented the components of all modern cryptographic algorithms, he believed that his strong ciphers are impractical for communications, due to the susceptibility to errors. The error-correcting codes invented by Hamming have solved this problem, as established later by Horst Feistel at IBM.
That is a fictional character in the novel Cryptonomicon by Neal Stephenson.
Moreover, the action of the novel is in WWII, many years after the public disclosure of the one-time pad by Vernam, which happened in 1926.
While the novel Cryptonomicon uses the term "one-time pad", it is very likely that this is an anachronism, because I have not seen any document from WWII or earlier that uses this term. The classified manuals of Military Cryptography and Military Cryptanalysis written by Friedman described it without using a special term for it, while Shannon, in 1945 and 1949 called it the "Vernam system", quoting the only non-classified source for it, i.e. the paper written by Vernam. I believe that the term "one-time pad" might have been coined during the Cold War to describe the ciphers used by Russian spies, who used random keys written on sheets of paper, which were destroyed after use. So in the beginning it was not a term referring to ciphers implemented by machines.
Before the classified work of Friedman from 1918, who cryptanalyzed documents intercepted in France in the final months of WWI, there was a certain Frank Miller who has described a kind of one-time pad in 1882.
However, what Frank Miller has written did not have any influence on cryptology. Moreover, his choice appears to have been just a lucky guess, which was not based on any experience in breaking ciphers or on any mathematical theory.
Leo Marks' memoirs, _Betweek Silk and Cyanide: A Codemaker's War, 1941-1945_ about his time as chief cryptographer of SOE, frequently discusses two major ideas of his that he pushed regularly: Worked Out Keys (WOKs) and Letter One-time Pads (LOPs). His heavy use of that acronym in his book written in 1998 is pretty strong evidence to me that he at least used that term of art during the war. Now, at least as presented in his memoirs, he was mostly isolated from the main cryptographic efforts of the war, so it seems unlikely that e.g. Meredith Gardner and the Venona Project would have encountered his use of the term, so I think that strongly suggests that the term pre-dates WW2 unless it was a simultaneous coinage.
That would make "one-time pad" a British term, which is consistent with the non-existence of this term in the early American documents.
Even if "one-time pad" had been used by the British during WWII, that would still make its use in Cryptonomicon inappropriate, because there it was used by an American.
Thanks for pointing to Leo Marks' book. I have just browsed it and it is weird how unfamiliar he was with the previous cryptographic literature, despite being a trained cryptographer.
According to his memoirs, Leo Marks had great difficulties in rediscovering the "letter one-time pads", in order to replace the "digit one-time pads", which were inconvenient for Morse transmission.
Not only the solution to his problem was clearly explained in Vernam's article from 1926, which had been published in both the Transactions of the A.I.E.E and in the Journal of the A.I.E.E, which were journals important enough to be available in various British libraries, but the solution searched by him was also explained in various popular publications, even in one of the novels written by Jules Verne almost a century earlier.
Anyone familiar with the history of cryptography and with the various kinds of ciphers used in the past would have thought instantaneously to the correct solution for implementing the desired "letter one-time pad" (i.e. by addition modulo 26 of the numerical positions of the letters in the alphabet). Also, had they been well aware that good one-time pads are unbreakable, they should have easily realized that the double encryption with a codebook followed by a one-time pad is useless.
From his memoirs, it appears that his knowledge of cryptography, at least in the initial part of WWII, was much inferior to the content of the manuals written by Friedman, which were used for training the American cryptographers, although it appears that in time, after gaining experience, he has become good enough.
As he presents himself in the book, at least, he was basically a dilettante, one of many who got sent to the GC&CS at Bletchley (the people who get described in Kahn 1991 or Winterbotham or Calvorcessi in their explanations of who worked at Hut 6 or 8 as 'linguists, mathematicians, people who wrote crossword puzzles' types). He flunked out of GC&CS, however, and only got the job with SOE by the skin of his teeth (the general in charge of SOE wanted him to decrypt an actual operational message but forgot to give him the key- the general wanted to see how fast he was at doing the double transposition cipher compared to a clerk, but instead Marks over the course of a day's work cracked the actually sent operational message without the key, which impressed/scared the general far more and got him the job).
But he had just rapid wartime training of learning by doing, and he was largely by himself as the only cryptographer in SOE, at least as he depicts it, so it highly plausible to me that he missed a lot of things that were widely known in the broader cryptography community. (It was because he was so isolated from the rest of the British crypto community that he ended up allowed to write public memoirs, I suspect. Wiki says he wrote it in the early 1980's and wasn't allowed to publish until 1998.)
NSA are purple team (both red team and blue team), so they do defense aswell as offense. They need to sniff plaintext aswell as protect their own infra and IP with strong crypto standards like AES. The public also benefits from AES, often to the detriment of SIGINT efforts by the NSA, so there are caveats to this, and it's nuanced.
Suite A is fairly specialised. Nobody can publicly comment on specific use cases of course, but it's fairly well known that Suite B is used for most national security information, including TS/SCI. But yes, technically non-public algorithms exist and are in use, it's just not even nearly automatically applied to everything.
Cool! It would be fascinating to read more about the Suite A ciphers and compare them to known designs, see how they were inspired by, improved or differ. But I guess there's not much chance of that happening as they're classified, haha! :)
They're using fips like the rest of govt. i get that you don't know. That's fine, but when you start making up stuff to fill in the gap that's when you need to do a reality check.
Actually, two of the awards are for codebreaking in the (relatively) distant past, far before modern practices.
"Evelyn Akeley's... accomplishments during World War II were exceeded only by those of her students, who broke virtually every Japanese army code they encountered.
"James Lovell... 'the [American] Revolution's one-man National Security Agency.' His pioneering work as a codebreaker and codemaker gave cryptology a singular role in the emergence of our new Nation. Leveraging Lovell's decrypts, George Washington knew of the approach of a British relief force and was able to warn his French allies, thus enabling a decisive victory at Yorktown."
I see your point, but be careful of throwing the baby out with the bathwater on this one.
AES largely gained popularity in part by the NSA adopting it into the "Suite B Crypto" program. The NSA also helped strengthen and develop DSA & SHA during it's early years.
the NSA has always liked somewhat weak encryption. specifically, they would really like if encryption could be broken in very roughly 1 week to 1 year on a super computer.
NOBUS (Encryption which Nobody But Us can break) which is the rationale for DES weakness) doesn't make any sense any more today. Can the Americans spend a thousand times more on compute power than the Chinese today? Do they have uniquely intelligent mathematicians? No. So there's no point in popularizing any solutions that the US can break, because if they can the Chinese can break them too.
I think lots of people have this idea that everybody thought DES was fine and so the reality of attacks on DES was astonishing, therefore AES won't be any better we're just in the dark somehow. That's just not true, DES was known to be purposefully weak, good enough but not actually good. 56-bit keys and 64-bit block sizes - you can't brute force that with a computer you can buy from the store, but it's not ludicrous and clearly somebody with government money can do it eventually. AES makes those numbers enough bigger that you just can't break it this way.
Perhaps in the past, but the NSA are good at predicting trends. They know that something broken by a supercomputer today will likely be broken on a smartwatch tomorrow. Intelligence agencies, unlike most tech firms, need to futureproof their secrets for decades. While the NSA no doubt has an interest secret backdoors, I don't think they would today support mathematically weak encryption.
That doesn't make sense as an NSA target, because NSA's adversaries all easily clear that threshold. What NSA supposedly wants are "NOBUS" weaknesses: keyed vulnerabilities for which only they hold the keys.
A vulnerability isn't "NOBUS" just because it exploits an unpublished zero-day. Dual EC was NOBUS because exploiting it required a curve private key that presumably only NSA had.
I always assumed that the NSA 'hardened' versions of products or operating systems were a careful mix of fixes for things they wanted to be protected against, while still letting secret backdoors to be preserved.
I found that super interesting but couldn't find anything online about him, not even the era (Cold War, 911, recent). If anyone has any links or literature about things he did I would very much appreciate it.