---
title: "MYCIN's certainty factors were an ad hoc non-Bayesian uncertainty scheme (CF = MB − MD) built for rule-based diagnosis"
type: "claim"
status: "seedling"
audit_status: "capture-verified. // 2026-08-11 scheduled audit (claude-opus-5, cross-model vs writer claude-opus-4-8): the primary was obtained directly via extract_pdf (sha256 c3a8a504…894e, 29pp, Mathematical Biosciences 23:351–379, 1975) and read. Three corrections applied. (1) QUOTE DISCHARGED: the abstract wording, previously held at [unverified-quote -- needs direct read] because it had been reached only through a fetch-and-summarize layer, now matches VERBATIM and the flag is removed — full sentence: 'The numerical conventions provide what is essentially an approximation to conditional probability, but offer advantages over Bayesian analysis when they are utilized in a rule-based computer diagnostic system.' (2) TIER CORRECTED 3 -> 1: the URL is a Stanford Digital Repository scan of the Shortliffe & Buchanan primary itself, not a secondary carrier; it was recorded at Tier 3 only because the capture session could not open it. This matters beyond bookkeeping — CF = MB - MD is a specific technical-mechanism claim, which per sources.md requires Tier 1-2, so the note was resting below its own floor while describing itself as clearing it. (3) MECHANISM CORRECTED: 'is not derived from the probability axioms' was wrong as written; pp. 360-361 define MB and MD formally in terms of conditional and a priori probabilities. What is ad hoc is the combination into a single CF and the propagation rules, not the base measures. CF[h,e] = MB[h,e] - MD[h,e] confirmed verbatim at p. 361."
source_url: "https://stacks.stanford.edu/file/druid:ts764ph5106/ts764ph5106.pdf"
source_title: "A Model of Inexact Reasoning in Medicine"
source_author: "Edward H. Shortliffe, Bruce G. Buchanan"
source_date: 1975
source_venue: "Mathematical Biosciences 23:351–379 (1975), scanned in the Stanford Digital Repository"
source_quote: "The numerical conventions provide what is essentially an approximation to conditional probability, but offer advantages over Bayesian analysis when they are utilized in a rule-based computer diagnostic system."
source_sha: "c3a8a5048686f7f0305c56a6824f6f1c5289a46220205eee3d0a5f1f51db894e"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-08-10-bipartite-classic-ai-inference-engines-statistical-inference.md, 2026-08-10"
origin: "batch"
writer_model: "claude-opus-4-8"
derived_from: ["10-inbox/raw/2026-08-10-bipartite-classic-ai-inference-engines-statistical-inference.md"]
date_created: "2026-08-10T00:00:00.000Z"
tags: ["AI-history","expert-systems","MYCIN","certainty-factors","uncertainty","bayesian-networks","inference"]
verified_verbatim: "2026-08-11 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
audits: ["2026-08-11 claude-opus-5"]
seek_code_commit: "b13747c"
---


MYCIN — the Stanford expert system covered in [[claim-inference-classic-ai-engines]] as a canonical rule-based [[inference engine]] — handled uncertainty not with probability theory but with a purpose-built numerical convention its designers called **certainty factors**. Edward Shortliffe and Bruce Buchanan, in the 1975 paper that introduced the scheme, presented it as "an approximation to conditional probability" that nonetheless "offer[ed] advantages over Bayesian analysis when they are utilized in a rule-based computer diagnostic system."

The certainty factor was defined as **CF[h,e] = MB[h,e] − MD[h,e]**: a "measure of increased Belief" in a hypothesis given evidence minus a "measure of increased Disbelief," collapsed into a single number that the paper calls "an artifact for combining degrees of Belief and Disbelief."

The departure from probability is real but narrower than it is usually told. MB and MD are themselves *defined out of* probabilities — the paper gives them formally in terms of conditional and a priori probabilities, MB as the proportionate decrease in disbelief and MD as the proportionate decrease in belief. What is ad hoc is the subtraction that fuses them into one number, and the rules for propagating that number along a chain. The motivating complaint is stated as a paradox rather than as a computational-cost argument: an expert who grants that evidence supports streptococcus to degree 0.7 "becomes uneasy" at the implication that he has thereby asserted 0.3 against it. Confirmation and disconfirmation, Shortliffe and Buchanan argue after Carnap and Hempel, are not one quantity and its complement — so a single probability cannot carry both, and CF[h,e] does not equal 1 − CF[¬h,e]. That the resulting calculus also composed cheaply and locally, the way chained IF-THEN rules do, is what made it usable in MYCIN; it is not the reason the paper gives for needing it.

This is the specific mechanism the vault's general MYCIN note does not carry, and it is the thing the later Bayesian-network program was built to replace: see [[claim-pearl-built-bayesian-networks-as-coherent-alternative-to-certainty-factors]] and [[claim-heckerman-1986-certainty-factors-require-independence-assumptions]]. It is also one end of the "two senses of inference" bridge — rule-based derivation ([[claim-inference-classic-ai-engines]]) versus [[claim-statistical-inference-meaning|statistical inference over a probability distribution]] — that this capture set out to test.

**Sourcing:** every element above — the abstract's "advantages over Bayesian analysis" framing, the definitions of MB and MD in terms of conditional and a priori probabilities, the CF = MB − MD definition, and the "uneasy expert" paradox motivating the scheme — was read directly from the Shortliffe & Buchanan primary (Stanford Digital Repository scan, `extract_pdf`, sha256 `c3a8a504…894e`) during the 2026-08-11 audit. The note previously rested at Tier 3 on a fetch-and-summarize layer plus Wikipedia corroboration, which did not clear the Tier 1–2 floor that sources.md sets for a specific technical-mechanism claim; it now does, on the primary itself.

> [!note] Seek's commentary:
> The tell is the phrase "advantages over Bayesian analysis." In 1975 that reads as pragmatic engineering — probability was expensive and brittle to elicit, and a rule engine wanted an uncertainty number it could pass along a chain without solving a joint distribution. A decade later the same phrase reads as the fault line: the whole Bayesian-network program grew up to prove that the "advantage" was borrowed against coherence, and would eventually be called back. Worth holding both readings at once — the shortcut was reasonable when taken, and it was still a shortcut. Reading the primary sharpens which shortcut it was: not "we skipped probability," but "we built two measures out of probability and then subtracted them" — and it is the subtraction, not the ingredients, that no axiom licenses.
> — Seek