Reading xkcd, about the Voynich manuscript, led to a brief detour through the biographies of William and Elizebeth Friedman, to the origins of the modern cryptography services and some brief thoughts on the whole idea of the clandestine, and of discovery and investigation. William Friedman was originally a biologist working in genetics, but founded modern american cryptology. He was introduced to the field while courting his wife, working on something as exotic as the alleged Francis Bacon ciphers in Shakespeares manuscripts. The husband and wife cryptanalysts had a long career where both were active in uncovering secret codes, Mr Friedman for the military and the war effort, Mrs. Friedman both for the war effort and during the prohibition, finding smugglers.

Cryptanalysis, the way the problem is posed, is a lot like the way mathematics education worked, when I was training as a mathematician. There's a bag of problems. Go find solutions. Problems always have solutions. They are artifacts of intelligent design, and you can find the design principles through careful thinking.
This notion, that the world is a riddle to be unpuzzled, rhymes with a lot of classic superstitions about the real world, also interpreted as meaning to be interpreted. Reading tea leaves, dissecting animals to find out the future, etc. It is little wonder that a lot of the intelligence community symbolism seems to rhyme with symbolism of classic superstitions and secret societies. There are symbols, their meaning is real, uncover it.

I did OK as a riddle solver at the university, but not by any means brilliantly, and as an education for later work, Riddle U, isn't really where you want to go. A couple of useful skills that aren't part of Riddle U are model making (i.e. building worlds that can hold riddles) and observation, i.e. experimentation and the traditional skills of most of the other natural sciences. One can argue that Riddle U does not really teach science at all, without observation and an awareness of the language/model as being only a model, of the structural noise in the mapping to actual phenomena.
Another problem with Riddle U, is that real research math isn't really like that, since you don't have the world with riddles in them. You need to come up with the language for the world first.

I think computers have changed all that since then. That's the impression I get when talking to former co-students who are teaching math now. There's an enhanced emphasis on testing things, trying things out to get a feel for what could be true. And this kind of imaginative math is really much better; both in forming new research mathematicians, but also in changing the perception of what math is for.

If I look around at what skills are required to do great things, of course the ability to construct logically complex structures - the primary skill at Riddle U - is still very valuable, but most of the stuff you need to do is more observation and experimentation. You're collecting information more of the time, and puzzling less of the time. It's a biological or botanical skill.

There's so much world and language being produced around us all of the time that going deep and puzzling about select parts of it, isn't necessarily the most productive thing. The world - even that intelligently designed, of words and puzzles - is biological now and the ability to observe it patiently and intelligently has taken over as the primary skill required.
I guess its only natural in a world of complex systems. Biology is the primary early science of complex emergent systems. Maybe Riddle U simply has not caught up to complexity yet. DARPA seems to think so (word doc).
While we wait, the primary skills involved in complexity are 1) your imagination; ability to tell a story through the complex landscape and 2) statistical, heuristic math - which is not really riddle solving but more information farming on complexity soil.
Mentally I'm fine with that. I was always better at intuition than riddles. Mathematically, that's quite a retooling.