For a very long time I've regarded Racket as a purely educational Scheme dialect not particularly useful for real work. It was only a few months ago that I actually bothered to look at it's (amazing!) documentation [1] and realized how wrong I've been. It's incredible what are the Racket guys doing and what they've already achieved. The Typed Racket [2, 3] is "only" a cherry on top.
What I see is that in part the transition from MZscheme to Racket has meant that a new generation of Lisp oriented academics are in PLT leadership roles - this is something Felleisen has long promoted in his own work (e.g. listing junior authors first on joint papers since everyone already knows who he is).
In other words, Racket is now being driven by research into computation and the earlier generation's area of emphasis - the pedagogy of computation - is being treated as a largely solved problem. This is not to say that PLT is no longer pursuing pedagogical projects or that in the past there wasn't significant discovery and invention within computation, only that the How to Design Programs project offers a mature and battle tested framework for teaching introductory programming and there is not a lot of reason to revisit it.
One of the things that attracts me to Racket - beyond the fact that it is a Lisp - is that the documentation is full stack. I was aware of the paper because there is a link within the Racket documentation of the contract system. The payoff from reading a paper that is specifically organized around the language at hand is intellectual continuity and when the language is a Lisp that continuity extends all the way from implementation to the lambda calculus.
[1] http://docs.racket-lang.org/
[2] The Type Racket Guide: http://docs.racket-lang.org/ts-guide/
[3] The Type Racket Reference: http://docs.racket-lang.org/ts-reference/