Struck by the Serpent

Fallen from grace, is the Serpent Cipher,
one part of a system, which I require.
A beautiful program could I once see,
but implementation wasn't to be.
As I read its single, simple paper,
interest of mine began to taper.
What I found I didn't like:
a serpent, ready to strike.
I was struck by the realization
it's not an interesting creation.

I knew well the kind of code I wanted,
but by boring code would I be taunted.
The APL needed iteration,
and Lisp couldn't use table creation,
since Serpent requires I study a tree;
it's a boring game of P, or NP.
What I found I didn't like:
a serpent, ready to strike.
I was struck by the realization
I won't understand this machination.

The tree becomes a writhing mass of snakes,
suffocating what forms the code then takes.
Know cryptographic code's not fun to write,
efficiency becomes primary plight.
The math, this code, so heavily constrains,
that the hacker, I, it heavily pains.
What I found I didn't like:
a serpent, ready to strike.
I was struck by the realization
at least I'm done with this foundation.

I had more fun implementing SHA than Serpent.  With SHA, I had room for creative designs and naming;
I was able to take different approaches to implementation also.  With Serpent, I'd relatively few of
these choices; even the naming was for the most part obvious.  Serpent also has optimized methods of
performing its substitutions, meaning I saw no point in a Lisp implementation, as such Lisp couldn't
reasonably have the tables and derive such optimized methods through metaprogramming; such searching
is much less fun than metaprogramming.  Even despite this, I wrote most of an APL Serpent before the
key scheduling posed an issue that would have me writing iterative code, at which point I decided to
cease with it; the APL needn't be practical, and my table derivation code was pleasant, but oh well.
The Serpent Cipher is fallen from grace for the reason it was a losing candidate for the AES system.
.