Summary
The 1987 collapse of the expert-systems industry. LISP machines lost to general-purpose workstations; expert systems proved brittle and expensive to maintain. Japan's Fifth Generation Project (1982-92) failed to deliver. A second cycle of hype and reversion.
Objectives
- 01Explain the LISP-machine market collapse.
- 02Name two expert systems that failed to sustain their promise.
- 03State the lesson practitioners drew and how it shaped the statistical era.
The Lesson
The market cause
Symbolics and LMI built specialized LISP hardware. Sun workstations running LISP well enough at a fraction of the price ended the market by 1987.
The paradigm cause
Expert systems required extensive knowledge engineering, could not learn, and degraded silently when the domain shifted. Maintenance costs exceeded gains. XCON, once a success, became a cautionary tale.
Key Ideas
- Hand-coded knowledge does not scale to the world.
- The second winter drove the field toward learning from data.