[Cryptography] Random permutation model for encryption as a teaching tool?

[Stephan Neuhaus]

Dear list,

thanks for the many responses here. I need to emphasise though that the 
students will not be asked to analyse or construct block ciphers, so the 
inner workings of a cipher are wholly immaterial.  Some responses have 
been along the lines of "if you explained *this* particular way that 
ciphers are constructed, they'd have a better understanding of *that* 
property". These may be true, but they are not what I've been looking 
for. I've rather been looking for a mental black-box model of how 
ciphers work. I probably should have explained that better.

 From the responses, here is what I got so far:

* Use "pseudorandom functions" a la Lindell & Katz instead (in 
"Introduction to Modern Cryptography") (Thanks, Jon)
* Don't talk about permutations that aren't good for block ciphers -- if 
they're random (or well-chosen pseudorandom), they'll be good for crypto 
(thanks Peter, John, but see also Thierry's comment below)
* A topical Dilbert cartoon 
https://me.me/i/tour-of-accounting-over-here-we-have-our-random-number-11012093 
(thanks Peter)
* Don't talk about "permutations of b-bit blocks" because it will make 
the students think that the bits in the block will be shuffled around. 
(thanks, John)
* A crypto *system* (as opposed to just a cipher) should use a different 
permutation for each block in the message, cf. IVs, chaining modes 
(thanks, John).
* Finding a good heuristic for the selection of 2^n permutations out of 
the (2^b)! possible ones is precisely the problem statement for a "good" 
block cipher (thanks, Thierry)

Very useful comments, thank you all very much!

Fun,

Stephan