MDICK project to secure America's borders with cutting-edge white-whale-related technology [updated]

Not a dead dingo's dick

The consensus around the Riddled office is that no-one gives two tugs on a dead dingo's dick about CYC these days. Also that it is immature to reap so much amusement from the mismatch between the ambitions and promises of Artificial Intelligence researchers vs. their actual accomplishments. But I maintain that the story of Douglas Lenat and his three-decade-long quest to perfect CYC -- a software expert system with common sense and natural-language capabilities -- is the Great American Monomyth, Ahab-&-the-Maltese-falcon-white-whale updated for today.

CYC's genesis was in 1984, and comparisons were rife with HAL-9000, though without the slight flaw in HAL's character. Lenat wrote a glowing account of CYC's prospects in 1995:

CYC: a large-scale investment in knowledge infrastructure

Not a chain of logic

Here is a droll account of evaluating CYC's performance a year earlier . By then CYC could draw on a database of 500,000 logical propositions (up from the previous 2,000,000 propositions) to concatenate into whatever chain or cross-linked fractal mesh of geometrical logic was needed to address a given question. Ambitions for CYC's possibilities had grown to searching a 20-image database of verbal captions, for suitable illustrations. The evaluator's list of natural-language questions went quickly by the board.

Lenat was much-run-after in the 1990s by technotopian magazines, as a reliable source of wide-eyed wonder:

I think it's conscious now [...] In the coming year, CYC will be out on Macintoshes and 486s.

HAL was a general artificial intelligence, and Cyc is the closest thing that exists in the world to that kind of general artificial intelligence [...] I believe [speech input and output] will be a small number of years -- possibly two years, which would be nice because then it would actually be during 2001.

The obsessive quest and the need for corporate / government research money continued into the mid-2000s, with 'CYC' becoming 'Cyc', though without perceptible change in its progress reports and objectives. Presenting questions to the oracle still required human interpreters. Reflecting the Zeitschmerz or the Weltgeist or whatever, Lenat's rhetoric put more emphasis on "citizen scientists" and the "blue-collar philosophers" programming Cyc with common sense rather than pointy-headed ivory-tower stuff. He became an inspirational speaker for Singularity conferences:

Cyc’s success could trigger the Singularity, so that’s a topic to which he’s given some serious thought, which he will share with us in this session.

"So where is CYC now?" people ask. Let's pretend people do ask that, instead of "Why are you all loonies?" and "Where is the Jesus tentacle porn we were promised?"
Here it is:
Ah, National Security and the GWOT. Is there any discredited dead-end technology it can't pour money into? It is only a matter of time before Captain Ahab receives a $2.10⁶ grant from the Department of Homeland Security for MDICK.

In other news, the current project for Riddled Research Laboratories is 'BuyCYC-l': a common-sense Expert System that can determine whether a given Expert System is a waste of money, brains and time, and whether the chief software architect is a delusive dingbat. Sadly, recent bilateral agreements with Australia prevent us using the acronym wombat, for fear of the offense that would cause for our marsupial-loving friends across the Tasman. In return they have promised not to use "Taumata­whakatangihanga­koauau­o­tamatea­pokai­whenua­ki­tana­tahu" as an acronym for anything.

BONUS 1994 explanation from Lenat:
"We're not interested in what botanists know about plants, but what chemists and musicians know. That's common sense."

This worked so well for English history.