In the 1970s to early 80s, these two AI programs by Douglas Lenat pulled off quite the feat of autonomously making interesting discoveries in conceptual spaces. AM rediscovered mathematical concepts like prime numbers from only first principles of set theory. EURISKO expanded AM's generality beyond fixed mathematical heuristics, made leaps in the new field of VLSI design, and famously was used to create wild strategies for the Traveller space combat RPG, winning national competitions two years in a row, even across rule changes to stymie it, before semi-voluntarily retiring. His magnum opus Cyc was originally intended to be a knowledge assistant to EURISKO's discovery engine.
These first two programs have intrigued the symbolic AI scene for 40+ years, with their grand claims but few eyewitnesses. While AM was technically available to fellow Stanfordians at the time, Lenat kept the source code to EURISKO close to his chest. Papers written about them carefully avoided technical implementation details. Lenat said he didn't retain any copy of the programs, when asked in recent decades, nor have any copies of AM carried publicly into the present.
It all started with this basic musing in a discussion with aindilis:
2023-11-24 22:50:03 <White_Flame> I also find it kinda funny that he shows where the code was in the school computer system and encourages people to download & modify it. Meanwhile, just a few years later he seemed to become super protective of his software.
Finding that quote again in Lenat's thesis, page 337:
Said running AM program is stored at SUMEX, on directory <LENAT>. From Interlisp, one need only load in the file <lenat>LT. This in turn will load in three files: TOP6, CON6, and UTIL6. So if you want to steal AM, take all four files!
While the name "AM" has always been completely unsearchable, I thought to try those particular filenames when lo and behold...
We poked around SAILDART and voila, there's user DBL himself (Douglas Bruce Lenat) and all his files from his Stanford days, including AM, EURISKO, and Traveller files! He noted an index of his files in WHERE[RDG,DBL]. AM has its own project area, but EURISKO and Traveller-related files are in there as well, and aren't as well delineated as the main AM files from the thesis, requiring further spelunking.
UPDATE 2024-01-02 - Roughly documented & topically grouped the DBL areas here in the EURISKO github wiki.
HUGE UPDATE - User seveno4 has run EURISKO in Interlisp Medley "with almost no changes"!
The DBL files on SAILDART were password protected until Dr. Lenat's passing, which explains why those have never turned up in searches before. I agree with respecting the privacy of the living since these were personal files, but also the benefit of sharing these afterwards. The value of this long lost seminal work cannot be overstated to AI historians.
We have not yet attempted to run any of this, but coincidentally I've already been loitering around the Interlisp revival folks, so we'll see what's possible. (I'm mainly a Common Lisper.) The coding style is of that era, dense and often uncommented. Just like SHRDLU, I doubt the same code will be extended beyond just getting it running, without total rearchitecting. But it certainly looks like it might be complete!
I'll also see what it takes to get the AM parts of the Stoyan documents scanned and compared to these versions. Unfortunately, EURISKO does not appear to be in that set.
Massive thanks to Bruce Baumgart for creating and hosting SAILDART, as well as Herbert Stoyan, the Computer History Museum, and all the other archivists out there. You never know what from the present will become important lost history.
UPDATE Github repo for AM: https://github.com/white-flame/am
and EURISKO: https://github.com/white-flame/eurisko
Discuss on Discord
Discuss on Hacker News