[Cryptography] how to detect breakage -- lures etc.??

[John Gilmore]

> The Enigma machine was /almost/ unbreakable.  It took tremendous
> cleverness and enormous resources to break it.  A layer of modest
> superencryption would have pushed it far, far out of reach of the
> codebreaking technology of the day.

It's well worth reading the books about the breaking of Enigma, like The
Hut Six Story by Gordon Welchman.  The key lesson is that breaking
Enigma on an ongoing basis required breaking the very early versions,
which were the least secure, and understanding how the Germans used them.
The German military produced many variants of Enigma throughout the war,
each one stronger than the previous ones.  After a period of being able
to read German traffic, everything would "go dark" when a new version of
Enigma was deployed.  Then the teams would feverishly work on figuring
out what had been changed, for months, until occasional messages could
be read, and eventually almost all messages.

The small Polish team that first broke Enigma, broke the version with
three rotors, and was able to read messages "almost every day" for most
of a year, until December 15, 1938, when the Germans introduced two more
rotors with different wiring, any three of which could be inserted into
the machine in any order.  The Poles were able to figure out the wiring
of the two new rotors, but unable to put forth the effort to break the
whole system, and their country was about to be attacked and overrun.
So they handed over their work product to the UK and France, including
working Enigma machines.  The UK was able to build systems and
procedures that could break the 5-rotor-in-3-slots Enigma, and begin
reading German messages.

Throughout the war, different, stronger variants of Enigma were used by
different groups (e.g. the German Navy, the top generals, diplomats,
etc).  For example, a fourth rotor slot was added to some; and a
plugboard was added to provide additional scrambling.  Each variant had
to be figured out.  Knowing how the basic machine worked, and what the
habits of the machine operators were, was essential in reducing the
combinatorial space to something barely solvable.  And even solving
these required inventing and evolving brand-new code breaking machines
unlike anything ever built before.

The high order bit is: Break the easy, early systems; develop expertise
in the communication habits of the enemy; and that gives you "cribs"
(suspected plaintexts or somewhat visible correlations) that enable you
to break the harder, later systems, which would otherwise be way beyond
your capabilities.

I suspect that this advice is still good.  Which is why inventing and
deploying cipher systems without also having a wiretapping, codebreaking
operation close at hand, is such a chancy business.

	John