The case for and against Arc is the same as for any other programming language: Does the syntactic sugar over the Turing Tarpit make the task at hand easier or harder. None of which is to suggest that I typically see anyone making a strong case for Arc these days.

I mean if anyone really was actually making any kind of case for Arc, they'd probably start with organizing it and promoting it in ways that would facilitate its inclusion as just another language in the batteries included distributions of Racket.