In my researches, I've found that machines such as the Symbolics LM-2, the
original LMI machine modelled after an LM-2, and systems such as the MIT
CADR were all microcoded machines. I'd think that this argues for a less
restrictive definition of LISP machines ;-)