The best way to do this is to probably add/extend classes, document, and implement along the way. That way anybody can jump in and help at any point, see what's been done, read the rationale and usage notes, and use it or help finish it.
I'm going to add some parsable comments to core1.s2 that I'll have auto-generated to a core layer reference in HTML.
Then, we need artsy people to start making styles, and artsy/programmer people to implement those styles in S2, and the total geek people to just review the classes and help implement incomplete parts.
Mart has a ton of stuff waiting to go into the core layer, but it's too much for me to handle... I have a half dozen emails from him in my inbox, all diffed against old versions, before we made certain decisions as to conventions, etc. He's terrible at sending me small changes, but I hope he'll get better.
Once the HTML reference is up, then all small changes can be proposed, discussed, and understood in this community first, before going into the core.