A long time ago I learned about information theory and saw some fascinating results obtained when examining encoding systems such as the English language and the genome. I was especially interested in the realtionship between efficiency of encoding systems and the probability of errors in transmission. IIRC these or the relationship between them were similar between English and the genome, but it was a long time ago.

Of couse similar considerations apply in software languages, if you look at the efficiency in encoding algorithms and the probability of bugs.

In both cases lowered efficiency in encoding information can allow for lowered probability of error. Think about golf-style Perl or assembly as compared to C#. Or even golf style Perl compared to Strict Perl.