Indeed, we don't know how the languages were selected -- certainly not by the paper's author, so you cannot fault him for this -- though I can guess:
- C++ needs no explanation.
- Ada was (is?) commonly used in defense systems. It was referred to as the "control group" by the paper's author.
- Some are related to Ada.
- Others seem experimental languages, part of a "ProtoTech" program for the design of prototyping languages. It seems natural to include experimental prototyping languages in an experiment about rapid prototyping!
I expect SML would have fared similarly to Haskell (or Miranda); at this level I would consider both to be more or less the same language. And it would definitely be interesting to see a Smalltalk and more modern languages!
> [the choice of languages] looks like a deliberate bias in language selection, and it makes it look like the conclusion was drawn before the study even began.
This is explicitly shown not to be the case. The experiment wasn't designed by the FP community but by ARPA and the Office of Naval Research; the review panel was unfamiliar with FP; none of the requirements was devised with FP or Haskell in mind. The paper's author is on Haskell/FP's side, but the review panel was not. Of all possible objections, we can safely disregard this one.
- C++ needs no explanation.
- Ada was (is?) commonly used in defense systems. It was referred to as the "control group" by the paper's author.
- Some are related to Ada.
- Others seem experimental languages, part of a "ProtoTech" program for the design of prototyping languages. It seems natural to include experimental prototyping languages in an experiment about rapid prototyping!
I expect SML would have fared similarly to Haskell (or Miranda); at this level I would consider both to be more or less the same language. And it would definitely be interesting to see a Smalltalk and more modern languages!
> [the choice of languages] looks like a deliberate bias in language selection, and it makes it look like the conclusion was drawn before the study even began.
This is explicitly shown not to be the case. The experiment wasn't designed by the FP community but by ARPA and the Office of Naval Research; the review panel was unfamiliar with FP; none of the requirements was devised with FP or Haskell in mind. The paper's author is on Haskell/FP's side, but the review panel was not. Of all possible objections, we can safely disregard this one.