I think anyone who is interested in a lisp has been positively influenced by Paul Graham, I certainly have.
You said this in the HN thread:
"I'm sure Arc is the wrong choice if I want to be as effective as possible in 6 months. If Arc is only the best choice if I want to be as effective as possible in 5+ years someone save me ;)"
The specific language is only one aspect of effectiveness - others include available libraries, an active contributing community, etc. I don't think there's reliable evidence that Arc will change enough in 3 years to fulfill your "effective as possible" goal.
Paul's last submission to his own forum has been over 5 years ago! And his last comment has been over 4 years ago. It would seem that he's not actively involved any more.
I was very excited to hear about Arc over 7 years ago, and I had high hopes that Paul might be the person to create a successful lisp and rally a great community. This has not happened, and I think it would be foolish to think this will change much in the next three years.
In the end, I ended up choosing Racket as my lisp - it made my short list to evaluate in part because Paul chose it to implement Arc :)
I'm certainly not trying to "save [you]", but I see many similarities between your quest for the most effective language and my own and countless others :) The language is very important, and concision is important, but you really need to evaluate the entire ecosystem to make a good choice.
I've mainly just been learning the language. Working through some tutorials, reading books/articles, coding, etc. Did a few Project Euler problems, and now I'm working through SICP.
I burned a lot of time looking for my (not "the") perfect language, and finally settled on Racket for the dynamic language and OCaml for the statically typed language.
Now I'm trying see which problems are best suited to each of them.
Clojure has a lot of good things, but I've never been able to get past the dependency on the JVM, lack of tail calls, etc., but it's pretty darn concise.
