[Cryptography] Leo Marks' 1998 talk about WW2 SOE code-making and breaking

[Peter Fairbrother]

On 22/01/2026 21:43, Kent Borg wrote:
> On 1/22/26 1:02 AM, Peter Fairbrother wrote:
>> he dances around […] the method of verbally giving agents distress 
>> codes which the giver couldn't remember.
> 
> I'm not sure what that means, but as I guess I am intrigued. Anyone have 
> more to fill in what cleverness we couldn't be told? (And more 
> description of the problem being solved?)

Hmmm, the problem: you are going to parachute into Denmark to teach some 
local agents how to do codes and operate wireless sets.

Part of what you have to teach them is their secure and distress codes - 
eg add a letter to the third letter of their indicator group, choose 
poem words which give a 16-digit second transposition key. These codes 
are individual to each agent.

Problems: you can't know what these codes are, in case you are caught 
and tortured.

you can't give the agent anything physical in case the Gestapo find it.

you can't say to the agent, make up something random for the first 
message and repeat thereafter, because the first message needs to be 
authenticated.


So how do you tell the agent what his secure and distress codes are?


Peter Fairbrother

I have been using "code" here when perhaps I should have been using 
"cipher" according to modern usage, but afaict the distinction between 
the words wasn't the same during WWII - even though encryption was 
called enciphering or just ciphering, cipher machines used "ciphers", 
but hand ciphers were called codes.