🚀 go-pugleaf

Just to add to your Lisp Machine history, there was a company which 
licensed MIT ZetaLisp from which they built a Lisp Machine called the 
KPS (Knowledge Processing System) which ran on Norsk Data super 
minis.  It had many advantages, such as multiport memory, it used 
stock hardware, supported some 2000 x 1600 pixel screens, allowed 
integration with conventional programming and processing environments 
(like using graphics optimized hardware and Fortran to do serious 
number crunching), large disk clusters, server architecture, ...

The company who did all the development work on the KPS was 
Racal-Norsk, a joint venture between Racal (a UK defense 
communications and sensors specialists (and founder Vodafone)) and 
Norsk Data (Norwegian computer company).  Various other individuals 
contributed to the development including Richard Stallman, Dick 
Gabriel,  and Jack Katsburg.  The good news about the KPS starts with 
the fact that the development team did a formal analysis and 
specification of MIT ZetaLisp (the root of Symbolics, LMI and TI lisp 
machine architectures) which had never been done before.  The good 
news continued with their success at building a working system.

The bad news is that the parent companies fell out of bed just as the 
development was brought to fruition (after some dozens of man-years 
work).  At which point the 20 odd core developers were all summarily 
fired and the company assets moth-balled.

In terms of LM history, I believe that the KPS was significant, if 
only for proving that the Lisp Machine architecture can operate 
successfully under non specialist hardware.  When Intel produce a 64 
bit processor, I think the ZetaLisp specification produced at 
Racal-Norsk could be a valuable reference for a developer wishing to 
produce a Lisp Machine for the next generation.

Needless to say, my memory of these events is fading, so I do not 
guarantee my factual precision, but I think the gist is correct.

-Peter Paine