Comments by "" (@grokitall) on "LISP: Lex Fridman's favorite programming language" video.

  1. if you watch the video, he says indirectly, that they start of with pure logic programming, which is slowish, then cache the results into lookup tables, which makes them trivial to turn into pure functions, which can then be refactored to extract more predicates and better data structures, which are then turned into an imperative programming language for efficiency, which happens to be in java. it is not that java byte code is fast, but that long chains of implicit inference are slow, and the multiple steps caching those inferences speed things up, not the java byte code. turning it into functions in a common imperative language just makes the results easier for non lisp programmers to use.
    1
  2. i think the point about development speed is due to the fact that one of the basic ideas in lisp is that it is easy to create new functions and data types, which makes it trivially easier to turn lisp into a system containing a superset of the language in the form of domain specific language extensions which get you progressively closer to the problem domain. this in general makes hard problems a lot easier, and near impossible problems plausible. this simply is not part of the mindset used with imperative programming languages, although it can be done in them. it is much faster to work in dsl's closer to the problem space.
    1