when/who proposed and developed L(R) as a model of ZF? In trying to understand L($\bf{R}$) better, I've not found an exposition of its origin as a model of ZF.  It was around before the end of the 1960s, and one might imagine there was an impetus for something "bigger" than Goedel's constructible universe L after Cohen's results on the CH; then again, Mycielski and Steinhaus proposed the axiom of determinacy (AD) in 1962, and that certainly has many desirable consequences in L($\bf{R}$), but hey, who knew (or intuited) such things early on?
So my question really asks for an overview of the chronology of what was conjectured/proved when, regarding key properties of the model.  (To clarify what I mean by key properties, I am thinking of things like "Dependent Choice holds" or "there's no nonmeasurable sets" or (under appropriate large cardinals) "AD holds", and so on.)
 A: Diving into what literature I could access, I have come up with an answer, in  "story" form, which more or less satisfies my inquiry.  (Perhaps it is an answer to a slightly different, or more well posed question.)  In any case, it makes sense (so far) to me...
As the 1960s dawned there must have been a certain air of anticipation in the mathematical world.
Godel's universe L of constructible sets had stood as a major advance for over two decades, an important inner model of the ZF
axioms.  It enjoyed the principle of Choice.  It showed the CH could not be disproved from the axioms of ZFC alone.
Yet there was a sense of things impending; many, even Godel, believed the CH was not "really" true (and might well be
a genuinely mathematical statement which was independent of ZFC.)  Dana Scott had shown that existence of a measurable cardinal
was incompatible with the axiom V=L. Impetus was thus given to the search for some bigger, better inner model(s).
New results and proposals now began to arise, such as Paul Cohen's invention of the powerful method of forcing.
Around the same time, Jan Mycielski and Hugo Steinhaus proposed their Axiom of Determinacy.  AD was quickly seen to contradict Choice,
but Cohen had shown both the CH and the AC to be independent of the ZF axioms; perhaps an accommodation might be found, short of the
full principle of Choice, for this intriguing new concept of Determinacy.
Ulam's old investigations of the measure problem had led to speculation about large cardinals (i.e. (2-valued) measurable) and
about any measure at all existing on the power set of the reals (real-valued measurable).  It was not long before Robert Solovay
showed the equiconsistency of existence of a measurable cardinal with a measure (indeed an extension of Lebesgue measure) on the power
set of $\bf{R}$.
Of course, a way to look for a model in which all sets were Lebesgue measurable would be to relax Choice, given that the AC
seemed to be essential in "cooking up" non-Lebesgue measurable sets.  Of course, some alternative principle strong enough
to yield a worthwhile theory of Lebesgue measure and integration would be desirable.  Around 1970, Solovay hit that particular
home run; he showed a model could exist (granting existence of an inaccessible) enjoying the principle of Dependent Choice (DC),
and with all sets Lebesgue measurable.  Though AD did not hold in the model introduced in that celebrated paper, he speculated
a large cardinal axiom might ensure that AD held in the inner model L($\bf{R}$). 
There were now opening up several paths through the still largely impenetrable forest of Set Theory, in addition to forcing. Large
cardinals; inner models; and determinacy.  Would any compelling results pull the various themes together, perhaps in some such bigger,
better inner model than L.  In the late 70s Solovay also showed the important principle of Dependent Choice was actually independent of
Determinacy.  What would it take, for DC to co-exist with the Axiom of Determinacy?
Then -- in 1984 -- Alexander Kechris got the needed breakthrough.  ZF + AD + V=L($\bf{R}$) implied DC.  As Ronald Jensen
has put it, this model is an analyst's dream.  All subsets of $\bf{R}$ in the model L($\bf{R}$) are Lebesgue measurable.  They enjoy the perfect
set property, so the Weak Continuum Hypothesis holds.  (There are no "intermediate" infinite sets, bijectable with neither the
integers nor the reals.) 
A final connection in this regard is that indeed (as Solovay had speculated) an appropriate large cardinal, a supercompact, suffices
to ensure AD holds in L($\bf{R}$).  That was later "brought down" to the existence of an infinite number of Woodin cardinals, with a
measurable cardinal above them all.
