The type system looks novel, but it doesn't provide controlled effects or safety as the system for Disciple language described in the "Type inference and optimization in impure world" paper. So it doesn't qualify as 'awesome' IMO

The docs are out of date. I read the Disciple papers about a month after finishing the thesis, and immediately began working out how to add "back" (regions were always intended) regions and effects.

Yeah, I have the same story. I have long wanted a "mutable" type system for my HNC project ( https://github.com/kayuri/HNC/wiki, http://code.google.com/p/inv/ ) - it was here on HackerNews about 1 year ago but got little to no interest. I asked on cstheory.stackexchange.com but got no useful answers. And then someone pointed my attention to DDC. I knew about it but I thought it was just an eager revamping of Haskell, just like many other eager haskells (speculative execution etc) out there.

The "newsig" branch is current development. That old stuff still has older syntax and semantics.