I don't think the conclusions of the article are actually supported by the arguments and evidence, and several of the technical claims seem pretty questionable (e.g. fast redefinition is not fast verification, entering a debugger is not guaranteed recovery, generated code is not automatically a validated library). With the same argumentation, Smalltalk (particularly Pharo), Racket, Julia, Elixir or even Forth would be equally credible candidates for "best programming language".