PARRY Parries Again - Reanimating the Famous Paranoid Chatbot
There were two versions of PARRY at Stanford in 1974, PARRY (aka old PARRY - deprecated) and PARRY2 (PARRY version 2). If you want to duplicate the PARRY/ELIZA interaction, you want to use old PARRY which I believe is what you’ve done. If you want to really want to be sure that you’re using old PARRY, use an old disk image from around say 1972 that doesn’t contain PARRY2.
I know all this because I’ve done a re-implementation of PARRY version 2. See
Interesting! Is PARRY2 also written in MLISP, and how did you get a copy? (Saildart.org I would presume, but.)
The original PARRY2 is written in MLISP (a LISP pre-processor - an Algol/LISP hybrid) and assembly language. Some of the code and data is present in the Carnegie Mellon University AI repository, but I got the full stuff at the SAIL DART archive. PARRY3 isn’t in the SAIL DART archive, because PARRY3 was developed on a different computer (SUMEX).
After further investigation, I’ve determined that Lars was in fact using PARRY2 re the ELIZAGEN stuff.
PARRY2 permits for the input parameter VERSION the value WEAK, MILD or STRONG. However, PARRY1 (old PARRY) permits for VERSION only WEAK or STRONG.
The May 1975 issue of the magazine “Analog Science Fiction - Science Fact” had
an article about PARRY (page 93, Turing Point - Thomas A. Easton).
See:
The article had the blurb:
The first “thinking” computer is now a reality. And it’s insane!