Some potential advantages of Arc
To me the (potentially) most interesting thing about Arc is that it takes a language family that is stuck and has plateaued in popularity and offers it a reboot along lines that are similar to other successful insurgent languages.
I'm not a Lisp expert, but here are my impressions:
1. Common Lisp is bloated, antiquated, fractured between implementations, and nearly impossible to improve except on a per-implementation basis.
2. Scheme is a much better language, but it's even more balkanized than Common Lisp, and the future of the language is being handled by a committee, which is rarely a good sign.
3. Many (most?) successful insurgent (i.e. not launched by a major corporation as a product strategy) languages have a single person who occupies the BDFL position. Smalltalk, Perl, Python, Ruby, and Erlang are all examples. Of course, counterexamples include Haskell and Scheme.
Rebooting Lisp with a new system started from scratch, having a single spec, and a single BDFL running the language is potentially a very good idea.
Sure, there's not much there right now, as everyone has complained, but there's enough there for a few motivated people (who can live with ASCII for now) to play around with, which is all you need for a version 0.01. If this is all there ever is, it obviously won't amount to much. On the other hand, if the community of interested people provide feedback and code and Paul starts building and/or blessing new libraries and code chunks for doing various tasks -- things like networking, data access, and, yes, internationalized strings.
People have talked about the social problems of Lisp for a long, long time. A lot of people (yes, I'm raising my hand here too) think this is the real reason Lisp has never taken over the world even as many of the ideas that Lisp pioneered become core components of more popular languages. What could be a better hack to fix a community with social problems than using a new language variant as an excuse for starting a new community?