Richard A. O'Keefe wrote:
> 
> It has been done.  You end up with a programming language that some
> people find elegant and highly usable, and some people find "simplified"
> to the point of making it hard to use.  This is already known.  What
> new information about the desirability or practicality of the change
> would be obtained by making this change to a Smalltalk compiler?

While it may not be new research in isolation, it seems like a natural
direction to pursue in conjunction with their other research.  It
seems particularly relevant to the Traits work, in which the use of
ivars is somewhat restricted.  So, I think it's a fine thing for them
to explore, but I also agree with the more conservative responders
that issues like encapsulation and specialized accessors need to be
addressed for it to have any hope of being a practical improvement
over Smalltalk as it is.

-Jesse