Certain languages are clearly better suited to some niches, though. I just learned awk, and, in retrospect, it's obvious that I've reimplemented a subset of it

Certain languages are clearly better suited to some niches, though. I just learned awk, and, in retrospect, it's obvious that I've reimplemented a subset of it in every other language I've used* . Sure, it's designed specifically for processing semi-structured textual data, but that's a surprisingly common situation. (Unix, man!) Since it's clearly a niche tool, it can dispense with a lot of boilerplate, and everything I need to know to use it fits in about 30 pages. (I've already written a couple useful three-line programs in it.) Likewise, if I have to calculate dependencies in a DAG, make immediately comes to mind. I think the real problem is languages that don't have clear mechanisms in place for co-operating with other languages (and their libraries). Without an FFI or other integration, you're stuck with a languages' weak points. It's too easy to get bogged down rewriting everything in Lisp (or whatever). * Granted, I don't like Perl (the heart of which is a sort of sprawling mega-awk).