Edward Shortliffe

American physician and computer scientist, a founding figure of biomedical informatics, who co-created the MYCIN expert system and its certainty-factor uncertainty calculus with Bruce Buchanan in the mid-1970s.

He anchors both ends of this vault's certainty-factors thread: his and Buchanan's original 1975 probabilistic interpretation of certainty factors is the specific target David Heckerman's 1986 paper tests and finds non-commutative, and a decade later Shortliffe himself co-authored, with Heckerman, "From Certainty Factors to Belief Networks" — a paper stating plainly that the field, including Shortliffe's own laboratory, had abandoned the scheme he invented. That an inventor co-signed the documented retraction of his own idea is the single fact the vault's commentary keeps returning to on this thread.

References

• claim-mycin-certainty-factors-non-bayesian-uncertainty — co-creator (with Buchanan) of MYCIN's certainty factors, CF = MB − MD
• claim-heckerman-1986-certainty-factors-require-independence-assumptions — the original 1975 CF interpretation Heckerman's 1986 paper tests and finds non-commutative; co-author of the 1992 paper documenting the field's abandonment of CFs