Chaocipher is a machine cipher that was undefeated for 90 years until the family of the inventor made public the author’s papers which described the algorithm for his simple two-disk machine.
This cipher is interesting for a number of reasons. The first is that Byrne published four texts with their cipher equivalents, amounting in total to over 10,000 letters, in his challenge to the world to evaluate the workings of his machine – and nobody succeeded. Not only that, but amateur cryptanalysts got nowhere near the solution.
The second interesting fact is that since the algorithm has been made public, the keys for only two of the four texts have been recovered. The keys used for the other two remain a challenge to be solved.
A third item of interest is that Byrne worked with a professional draughtsman in 1920 and produced a design for a fully mechanised enciphering machine with a keyboard. In the event this machine was never built because of its high cost, but we have available the original drawings though not a description of the machine. It remains a mystery at this time how the machine worked, and that is a puzzle still to be solved.
A final interesting fact is that professional cryptanalysts rejected Chaocipher as of no interest, but never expressed publicly their reasons. Notable amongst such critics was William F. Friedman, the leading US cryptanalyst of the 1930’s and 1940’s. But some current opinion regards the algorithm of Chaocipher as a clever enough idea to provide a strong, reliable and unbreakable cipher when implemented on a computer. This contention remains unproven one way or the other, and thus provides a challenge to the current generation.
I wrote an article on the Chaocipher algorithm, and how I solved two of the four Byrne texts, which you can read here.
There is a lot more information at the Chaocipher Clearing House.
Byrne demonstrated an early model to Friedman in 1922 when the latter was chief cryptanalyst in the War Department. We can safely assume that Friedman knew exactly how the system worked. At the same time Friedman was studying working rotor machines and so he knew the capabilities of existing and of developing equipment.
If Byrne’s principle had no merit, as Friedman later maintained in 1954, why didn’t Friedman say so in 1922 and also why didn’t he tell Byrne not to bother to send an improved model? Was there an ethic that in the cloak and dagger world of Cryptology you never revealed anything to a civilian? Possibly.
Perhaps in the technology of the 1920’s Chaocipher did have merit and that is why Friedman didn’t damn it. But then you would have expected Friedman to actively pursue it – which he clearly didn’t.
The likelihood in my mind is that in 1922 Friedman was told by a superior to humour a person of some influence who had a madcap idea. And Friedman did just that.
Once again in 1942 Byrne was asking Friedman to assess Chaocipher and once again Friedman was fobbing him off – this time with the request for a depth of messages. I see here the bureaucratic process at work. Friedman wasn’t interested (as witness his scribbled note to Lt. Hiser) but rather than be blunt about it he follows normal procedures and hopes the nuisance will go away -- which in the event it does!
Finally in 1954, when once again Byrne is taxing Friedman to acknowledge the merits of Chaocipher, Friedman has had enough. He tells Byrne to forget Chaocipher and focus on other things. That, as far as we know, is the last chapter on this topic in Byrne's lifetime.