Typed vs Untyped: performing a real experiment?
Prof Richard Connor, Strathclyde University
The long debate about whether static type systems are actually useful or
not, in terms of software productivity, quality, and maintenance, has
largely fizzled out with the acceptance of Java as a nondescript de facto
standard for software production. In its prime, the debate excited a great
deal of opinion, in the spectrum of reasoned argument, religious polemic,
and the burning of heretics by both sides.
What it lacked was any real evidence either way, by experiment or otherwise.
The essential difficulty is that forming a useful experiment over large
groups of progammers skilled in different languages was, and is, essentially
infeasible in terms of cost.
This talk outlines what we believe is a novel approach to the question,
which might lead to the gathering of some real experimental evidence about
this important, and undeservedly forgotten, question.