Thanks to Allen Wirfs-Brock's suggestion:
For more information on Gregor Kiczales' work on meta-programming and open
implementations see: http://www.parc.xerox.com/spl/projects/oi/default.html
which leads directly to the publication abstract and download links at:
http://www.parc.xerox.com/spl/groups/eca/pubs/complete.html#Kiczales:TUT95

I downloaded this document (Postscript) and converted it to PDF.  The
result is titled (surprise :-) "Open Implementations and Metaobject
Protocols", is 330 pages long, last modified Sep. 8, 1996, and appears to
be the draft Alan refers to below.  The PDF file is 827KB.  I'll be happy
to upload it if someone can tell me where (presuming Gregor doesn't mind).



Mike Wirth



"Alan C. Kay" <alank at wdi.disney.com> on 03/01/98 03:52:08 PM

To:   cohenb at gemstone.com, squeak at cs.uiuc.edu
cc:   wirth at almaden.ibm.com
Subject:  Re: "Open Implementations and Metaobject Protocols" (was Re:
      class, prototype, meta)





At 3:36 PM -0800 3/1/98, Bruce Cohen wrote:
-- snip -->
>Gregor was circulating early drafts to the object community, which may
>be why Alan Kay remembers reading a copy.
Yep, now that I vaguely recall, it very likely was a draft of someone's
that I perused.
What is so interesting about this subject area is the tantalizing notion
that by several mechanisms (many of them involving various kinds of
confinement) one should be able to get almost everything in a system to be
late bound without either great danger or significant performance hits.
Cheers,
Alan