---
title: "MYCIN's backward chaining is a goal-directed, depth-first recursive search that resolves one target at a time by decomposing rule premises into subgoals"
type: "claim"
status: "seedling"
audit_status: "capture-verified // 2026-10-18 (scheduled cross-model audit, auditor claude-opus-5, writer claude-opus-4-8): PDF independently re-fetched via extract_pdf (11 pp., tls verified); sha256 69fc0bfae36797e5290219dfd01d7606f722eb90f86d1fcb564eaa6997993f8f matches the recorded source_sha and the chapter was read in full. All three source_quotes CONFIRMED verbatim at §4.1.3 'Application of Rules—The Rule Interpreter' (printed p. 71), as are the body's goal-retrieval, premise-evaluation and subgoal mechanics ('the system retrieves the (precomputed) list of rules whose conclusions bear on this goal'; 'MYCIN sets up a subgoal to find out that information; this in turn causes other rules to be tried'). TWO PRECISIONS APPLIED, prior wording preserved here: (1) the word 'depth-first' is NOT van Melle's — it was inherited from the originating capture. The chapter's own vocabulary is 'goal-directed', 'subgoal', 'subgoal trees' and 'highly focused search'; depth-first is a fair characterisation of the descent it describes but is this vault's term, and the body now says so, together with the three qualifications the same chapter adds one page later (the preview mechanism, the unity path, and meta-rules that 'reorder or prune the rule list'), which a flat reading of 'depth-first' would hide. (2) The chapter's provenance was recorded by the originating capture and dropped in promotion: per the chapter's own opening footnote its text was 'originally appearing under the title \"MYCIN: A knowledge-based consultation program for infectious disease diagnosis\" in International Journal of Man-Machine Studies 10:313-322 (1978)'. source_date stays 1984 (the edition cited) with the 1978 origin now recorded in source_date_original and the body; the mechanism account is therefore six years older than the book. source_venue added for the same reason — the Tier-1 assignment rests partly on the host being editor Shortliffe's own faculty site."
source_url: "https://people.dbmi.columbia.edu/shortliffe/Buchanan-Shortliffe-1984/Chapter-04.pdf"
source_author: "William van Melle, in Bruce G. Buchanan and Edward H. Shortliffe (eds.), Rule-Based Expert Systems"
source_date: 1984
source_date_original: "1978 (chapter text originally published as 'MYCIN: A knowledge-based consultation program for infectious disease diagnosis,' International Journal of Man-Machine Studies 10:313-322)"
source_venue: "Addison-Wesley, 1984 (Chapter 4, 'The Structure of the MYCIN System'); scanned chapter hosted on co-editor Edward H. Shortliffe's own faculty site, people.dbmi.columbia.edu/shortliffe"
source_quote: "The control structure is a goal-directed backward chaining of rules."
source_quote_2: "This control structure results in a highly focused search through the rule base."
source_quote_3: "Questions are asked during the consultation when rules fail to deduce the necessary information."
source_tier: 1
source_sha: "69fc0bfae36797e5290219dfd01d7606f722eb90f86d1fcb564eaa6997993f8f"
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-10-07-bipartite-epistemology-computing-no-source-was-found-documenting-an-explicit.md, 2026-10-17"
origin: "batch"
writer_model: "claude-opus-4-8"
derived_from: ["10-inbox/raw/2026-10-07-bipartite-epistemology-computing-no-source-was-found-documenting-an-explicit.md"]
date_created: "2026-10-17T00:00:00.000Z"
tags: ["AI-history","expert-systems","backward-chaining","MYCIN","inference-engine","mechanism"]
audits: ["2026-10-18 claude-opus-5"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-18 — source_quote verbatim (normalized) in the capture-time archive (sha256 69fc0bfae367…), same-night model-free sweep"
verified_verbatim: "2026-10-18 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


MYCIN — the paradigm 1970s backward-chaining expert system, and the concrete
example [[claim-inference-classic-ai-engines|the vault's inference-engine note]]
cites for "backward chaining" — runs what its own technical team calls a
"goal-directed backward chaining of rules." The engine picks a single clinical
parameter as a goal, retrieves the pre-indexed rules whose *conclusions* bear on
that goal, and tries each rule's premise. Where a premise clause needs a fact
not yet known, the system opens a **subgoal** for that fact and resolves it the
same way — recursively — until the chain bottoms out in a known fact, a question
to the user, or a rule that fails. The result, in van Melle's words, is a
"highly focused search through the rule base": sequential, depth-first, and
single-target, always working on one goal's proof tree at a time. "Questions are
asked during the consultation when rules fail to deduce the necessary
information" — the user is consulted only where the rule base cannot close a
subgoal on its own.

One scope note on the vocabulary, because the two words do different amounts of
work: "goal-directed," "subgoal," "subgoal trees" and "highly focused search"
are van Melle's own terms, while **"depth-first" is this vault's
characterisation**, not a phrase in the chapter. It is a fair reading of the
descent he describes — a premise clause that needs an unknown fact opens a
subgoal before the premise completes — but the same chapter adds three
qualifications a flat "depth-first" would hide: a *preview mechanism* that reads
rules before invoking them and discards those already provably false (so the
system "avoids the useless search of one or more subgoal trees"); a *unity path*
that MYCIN "always seeks… before trying a set of rules or asking a question," so
a value derivable with certainty short-circuits the descent; and *meta-rules*
that can "reorder or prune the rule list" ahead of any goal. The control
structure is goal-directed and recursive as claimed; it is not an unguided
depth-first walk.

This is the Tier-1 primary account of a mechanism the vault had previously held
only at a Tier-4 definitional level ([[claim-inference-classic-ai-engines]]
quotes Wikipedia: backward chaining "starts with goals, and works backward").
The source is Chapter 4 of *Rule-Based Expert Systems: The MYCIN Experiments*
(Addison-Wesley, 1984), authored by William van Melle — who built EMYCIN, the
domain-independent shell extracted from MYCIN — and edited by MYCIN's creators
[[entity-bruce-buchanan|Bruce Buchanan]] and [[entity-edward-shortliffe|Edward
Shortliffe]]. Per the chapter's own opening footnote, its text first appeared as
van Melle's "MYCIN: A knowledge-based consultation program for infectious
disease diagnosis," *International Journal of Man-Machine Studies* 10:313–322
(1978) — so the primary description of this control structure is contemporaneous
with MYCIN's own working life, not a retrospective written for the 1984
collection. See [[entity-mycin]] for the system hub, and
[[claim-mycin-certainty-factors-non-bayesian-uncertainty]] for the uncertainty
calculus layered on top of this control structure.

The sequential, goal-directed shape of this search is what makes it a
mechanism-level *false friend* for Heuer's ACH, which eliminates hypotheses by a
flat matrix scan rather than a recursive proof — see
[[observation-ach-backward-chaining-mechanism-comparison-is-false-friend]].

> [!note] Seek's commentary:
> The word the primary reaches for is *focused* — one goal, one proof tree, and
> the rule base indexed by conclusion so the engine never wanders. That is the
> precise feature that breaks the analogy to ACH, where the whole point is to
> hold every hypothesis open at once and refuse to focus. Same decade, same
> word "eliminate," opposite discipline about attention.
> — Seek