Where the author mentions that it is related to his doctoral dissertation, Fexprs as the basis of Lisp function application; or, $vau: the ultimate abstraction[1].
I generally agree with alot the autor said and I will surly check out kernel.
About JITs
----------------
I agree that Scheme missed the JIT thing. Most Scheme compiler are AOT but Scheme would be nice to do research in JITs for dynamic languages. Now we have it in JS witch is more complicated. It would be nice if there would be a fast and small JIT for Scheme that is good for learning. is there something like that? I know Racket has something like that but Racket is much more then just a JIT for Scheme.
About Clojure
----------------
I think the author missunderstands Clojure (not just in this article). The "see the need (... of eval)"-statment was not about eval in general. Clojure has eval. The statmand was about having eval in ClojureScript (Clojure that AOT Compiles to JS) and there is not a priorety because its need is much less then the usfullness.
Clojure wants to be practical now and not be revolutionary. It never said it want to be ideal it just makes the best of what we have.
I've read about Kernel before, but this time I wondered if $vau is in any way related to 0 from Jot (the turing tarpit best apparently best used as a Goedel numbering).
there are certain similarities to other very-low-operator-count languages - the big difference is that with Kernel's fexprs, you can go from this handful of operators to a fully-featured language more expressive than Scheme, and do so with style. See the R-1RK - it builds up the whole language, sequentially, from the built-ins:
Where the author mentions that it is related to his doctoral dissertation, Fexprs as the basis of Lisp function application; or, $vau: the ultimate abstraction[1].
[1] http://www.wpi.edu/Pubs/ETD/Available/etd-090110-124904/