Well, Common Lisp or Racket are a lot faster than Python, too, and both are general-purpose languages which has many advantages over a language specific for numerical computing.
Like what advantages? Julia is a "general purpose language" too, every bit as much as any of those other languages. It all boils down to the availability of polished utility libraries, and I don't exactly think Common Lisp or Racket have an indisputable edge there.
I don't get this argument. Maybe you could elaborate. I tried for a long time to get used to Racket. I really like the idea of LISP. The reason why I started using Julia wasn't because it was good at numerical programming. I was because it was a much better GENERAL PURPOSE programming language than Racket.
Unless you do a lot of meta programming stuff, I don't find Racket to be advantageous to use. Julia syntax is much friendlier towards general-purpose programming. It is easier to do mathematics as well as string manipulation with Julia.
Multiple dispatch also makes it a lot easier to write code. Getting into the LISP heavy thinking of Racket is harder than jumping onboard Julia.
I find that I have been able to learn functional programming, meta programming and other things that make LISP/Scheme famous in Julia than I ever managed in Racket.