I don't see the point in this as a serious production language. As a project to just have fun, of course it is a cool idea.

Lisp systems moved to generic functions for a reason - Lisp dialects for some time used a Smalltalk-based message sending model, but it just isn't as flexible.

But if the desire is to get Ruby semantics with Lisp syntax, I think the much more reasonable approach is to run a Rubyesque Lisp on Rubinius (which the lead devs have already talked of doing). At least by doing that you get the advantage of piggybacking on an advancing VM and a language that already has libraries.

Establishing a new language is hard.

Disclaimer: I haven't tried Nu yet, just read the FAQs. Nu isn't about Ruby semantics, it's about Cocoa and Objective-C semantics. It's a Lispy/scripting way of interacting with the Mac OS programming paradigm. The author is also the author of the RubyObjc bridge, and Nu was a result of his frustrations with the "impedance mismatch" between Ruby objects and code and Objective C.

My first reaction is that I am surprised to see Lisp syntax with Ruby semantics. I would have expected things to be the other way around.

Yeah like Ruby syntax with Lisp semantics. Now that would be a sight to behold huh? although looking at it more closely it looks quite elegant. Worth a shot to learn.

Ruby has a fairly complex syntax. Why would you prefer it to Lisp's simple syntax? The nice thing about Ruby is message passing and dynamic method lookups, not its syntax.

Slightly off topic, but there was an interesting example on one of the documentation pages:

        % (set laws "Asimov defined #{(+ 1 1 1)} laws of Robotics")
        % (puts laws)
        Asimov defined 3 laws of Robotics

Are there any other standard ways to embed s-expressions in strings?