Here's a thought for a future project. The lexicon should be designed
so that in typical conlang text each letter of the alphabet gets used
an equal number of times.

English (and all other natlangs as far as I know) have oddly skewed
letter distributions. For example, _Moby Dick_ contains 117,051 of the
letter "e", but only 644 of the letter "z". A lot of perfectly good
letters, like {v,k,q,j,x,z} are seriously under-utilized. This seems
like a waste of perfectly good letters.

(Granted, there aren't too many "z"s in English since Polish used up
most of them after the invention of the printing press. I'm told
medieval printers either had to pay a premium to import the hard to
find "Z" slugs from Poland, or learn to do without them. The result is
the relative rarity of "Z" in English.)  ;-) ;-) ;-)

At any rate, having a language with balanced alphabet usage would make
it harder on cryptographers too.