> This is an answer from a scientist (or at least it looks like it, I don't know anything about Conal Elliott, not even if he actually is a scientist). He is t

> This is an answer from a scientist (or at least it looks like it, I don't know anything about Conal Elliott, not even if he actually is a scientist). He is the (or one of the two) inventor of FRP. > People like these think and speak on a level of abstraction which may be a little higher than what most people are used to. I don't care what 'level of abstraction' someone normally works on. When you're supposed to explain things to other people, you're supposed to explain it their level, not the level of someone who already knows FRP or who is also a programming language theorist/implementer. That was StackOverflow, not Lambda the Ultimate. > As a response from a scientist, his post is as good as you could hope, I think A lot of scientists are educators, and so they have to descend from their 'lofty towers of abstractions' to explain what a for-loop is to first year students. I have certainly seen many scientists, by whom I couldn't read and comprehend a single paragraph of their research paper, be able to explain things much more straightforwardly than him. > you still need a fair share of googling and reading to understand it, SO answers should try to be self contained. A general rule, but certainly doable for a question of the type "what does <term> mean?" > Denotational semantics is a tool like many others, A tool for prog. language theory and implementation. A tool for StackOverflow answers? Maybe, if you also consider answering SO questions of the type "what does this Scala program do" by converting it to an intermediate representation in lambda calculus, or answering "why won't this type check" by explaining with a type deduction in some sequent calculus, or by answering the question "what value will this formula give", by manipulating it as some algebra over some arithmetic signature. It might be beneficial to describe a language by some canonical semantic representation. For a SO answer? If you manage to write several paragraphs without even giving one introductory explanation/example/intuition about what it is, and then you catch yourself writing something like " If this description is gibberish to you, consult (a) Denotational design with type class morphisms,"... you should probably admit to yourself that you've become so accustomed to what is being asked for that you can't even imagine or sink down to the level of a person who has never been in touch with it.