E-recursion is a notion of generalized computability theory which seeks to extend computations to allow arbitrary sets as inputs. In contrast with e.g. $\alpha$-recursion, it disallows unbounded search: halting computations are witnessed by well-foundedness of related trees, while for many values of $\alpha$ well-foundedness of a given computable tree is $\Pi^0_1$ over $L_\alpha$. This leads to a really interesting behavior, with a number of odd properties.
I'm interested in the foundations. The standard definition of E-recursion is in terms of a set of schemes. Normann's original article on the subject is quite good, and presents this definition with solid motivation (another good source is the last part of Sacks' book, which certainly has much more material, but I've found it to be quite terse and not as good motivation-wise). However, definitions via schemes often (for me at least) leave some doubt as to whether they define the "right" notion of computability: how do we know we aren't "missing" more schemes?
Normann mentions, excitingly to me, that Moschovakis independently discovered essentially the same thing but via an inductive definability approach. This seems really interesting to me; unfortunately, as far as I can tell Moschovakis' paper never appeared (possibly because it was subsumed, modulo the definition, by Normann's).
My question is:
Does anyone have Moschovakis' text on the subject, or know what his definition was?