The Art of Artificial Intelligence
Feigenbaum stated the knowledge principle: a system's power comes from knowledge of the domain, not from the elegance of the inference mechanism.
Why it matters
The field formally changed its bet, from general search to narrow expertise, and that is what made the commercial boom of the following years possible.
The paper, given at IJCAI-77 and repeated at the National Computer Conference in 1978, put the phrase knowledge engineering into circulation; Feigenbaum's own list of publications says the 1978 version won that conference's award for the most outstanding technical paper. He describes the bottleneck: the knowledge must be extracted from an expert who often cannot state it. The field built such systems for the next eight years; the approach's limit, brittleness outside the described domain, showed in the second half of the 1980s.