The real world data is extremely limited, given there are only ~200 countries in the modern era, and the huge span of time over which the full impact of a constitution is seen.

It's incomparable to PLs, which have perhaps billions of case studies on which comparative analysis can be done.

Remind me how many PL's there are again.

Not GP here. I would estimate the number of programming languages is in the thousands, but the number of languages with multiple implementations and long-term maintained programs written in it is probably below one hundred. All the others do not contribute a meaningful body of real-world data for the purpose to learn from programming language design. But they might have contributed knowledge of about the design space of programming languages.

Actually the more I think about it, the more useful I do find the comparison between constitutions and programming languages.

Because there are probably only ~200 PL's with truly significant usage, and they tend to have a variety of purposes and be used in historically different contexts -- much like constitutions.

The same arguments that programmers have over declarative vs imperative is much like the arguments political scientists have over parliamentary vs presidential. Static vs dynamic type checking is much like civil law vs common law.

And there's similarly nothing "scientific" in proving that one programming language is "better" than another, but people have strong opinions, and experience will tell you which ones are good ideas to pick for a project and which ones are not.

By the way, I'm not sure you know what a case study is -- it's a report written by a person. Nobody's written billions of those. And you don't do comparative analysis on case studies -- a comparative analysis is a case study, where the cases are countries (or programming languages).