The AI Journal

Ockham's razor is introduced to undergraduates as a rule of scientific method: when two theories explain the same evidence, prefer the simpler one. The principle is usually stated in a Latin tag, entia non sunt multiplicanda praeter necessitatem, entities are not to be multiplied beyond necessity, and attributed without further comment to the fourteenth-century Franciscan William of Ockham. What gets left out of the textbook version is that this exact sentence does not appear anywhere in Ockham's surviving works. W. M. Thorburn traced the phrase's actual history in 1918 and found no citation for it earlier than the seventeenth century; what Ockham did write, repeatedly, was a narrower claim in a specific idiom, that plurality is not to be posited without necessity, deployed almost exclusively to settle disputes about God's power and the reality of universals. The razor now used to choose between rival physical theories began its life as an argument about what a Creator's omnipotence made metaphysically unnecessary.

The clearest cases are Ockham's attacks on realist ontology. Medieval philosophy before him was crowded with entities introduced to explain how thought and language connect to the world: universals existing outside particular things and somehow shared by them, relations as items distinct from the things related, a plurality of substantial forms stacked inside a single body to account for its different capacities. Ockham's standard move against each of these was not empirical but theological. God, being omnipotent, can produce directly whatever a secondary cause or an intermediate entity would otherwise be needed to produce; therefore positing that entity as a necessary feature of reality contradicts the freedom implied by omnipotence. If God could make Socrates white without an inhering, really distinct quality of whiteness doing the work, then such a quality is not a thing the world needs, only a convenience of grammar. This is an argument from divine power to ontological economy, and it recurs across Ockham's logical, physical and theological writing with a consistency that Armand Maurer's study of the underlying principles traces across nearly every corner of his output. The razor, on this reading, is downstream of a doctrine about what an unconstrained will requires, not a generalisation from successful scientific practice.

The case against real relations follows the identical pattern and shows the mechanism even more starkly. Scholastic philosophers before Ockham typically treated a relation, similarity between two white things, or a father's paternity toward his son, as a distinct item existing in addition to the two related things themselves, something the mind discovers rather than constructs. Ockham denied this by asking what such an item would be for. If God can make one thing similar to another simply by making both things the way they are, without inserting a third item called similarity to connect them, then similarity as a separate reality does nothing that omnipotence has not already accomplished without it. Marilyn McCord Adams's study of Ockham traces this argument through dozens of applications, from quantity to place to motion, and finds the same structure each time: whatever an all-powerful cause can produce directly, an intermediate entity is idle, and idle entities have no claim on existence. The argument is doing real metaphysical work here, but the work is licensed by a claim about God, not by a claim about explanatory economy in the abstract.

That doctrinal machinery is entirely absent from the sentence now offered as the paradigm case of scientific parsimony: Isaac Newton's first rule of reasoning in the Principia, that we are to admit no more causes of natural things than are both true and sufficient to explain their appearances. Newton's rule concerns causes of observed effects within natural philosophy; it says nothing about universals, relations, or the metaphysics of substantial form, and its justification is explicitly methodological rather than theological, resting on the claim that nature is pleased with simplicity rather than on any argument about what an omnipotent will can dispense with. The word razor did the work of collapsing these two projects into one continuous tradition, but the argument that licenses Newton's rule and the argument that licensed Ockham's are not the same argument wearing different clothes. Newton is making a claim about explanatory economy for causes of observed phenomena, defended by appeal to nature's regularity. Ockham is making a claim about ontological economy for entities implicated in what God's power requires, defended by appeal to divine freedom. The family resemblance between the two, prefer fewer things, obscures that one rests on a modal theological premise unavailable to a purely empirical enquiry, and the other does not.

This matters for more than antiquarian precision, because it changes what the razor is owed as an argument once it leaves theology. If Ockham's parsimony worked because omnipotence makes secondary causes and inhering universals gratuitous, that justification simply does not transfer to a debate between, say, two cosmological models fitting the same dataset. Nothing in the doctrine of divine omnipotence tells a twenty-first-century physicist why a universe with fewer free parameters is more likely to be the true one; the theological premise that did the work for Ockham has nowhere to attach. This is precisely the problem Elliott Sober's history and philosophy of the razor is organised around: having shown that most invocations of parsimony in the sciences cannot appeal to Ockham's own reasoning, Sober spends the bulk of his study trying to reconstruct, case by case, what could justify a preference for simpler theories in biology, physics and psychology, using tools like likelihood and model-selection statistics that have no medieval antecedent whatsoever. The need for that reconstruction is itself evidence that the razor's modern users inherited a slogan without its argument.

The most serious objection to this account is that it draws too sharp a line between the theological and the methodological. Even granting that Ockham's stated justification was doctrinal, the logical form of his move, do not posit an entity if the relevant effect can be explained without it, is identical to Newton's, and a shared logical form is arguably all that continuity across a philosophical tradition ever requires. Nobody thinks modus ponens is theologically compromised because a medieval logician first formalised it inside a treatise on the Trinity. On this view, insisting on the theological wrapper is a pedant's objection that mistakes the history of an argument's packaging for the history of the argument itself.

The distinction survives that challenge, but in a narrower form. Logical form alone cannot do the justificatory work that either Ockham or Newton needed it to do; in both cases, the bare instruction to prefer fewer entities requires a further premise explaining why fewer is more likely to be true, and it is precisely that further premise, not the imperative's grammar, that differs between the two traditions. Ockham had one: an omnipotent God does not need superfluous furniture, and finite intellects have no business positing what infinite power renders gratuitous. Modern science needs a different one, because it has no access to Ockham's, and the honest history of the razor is the record of philosophers building that replacement premise from scratch, repeatedly, without recognising that the original had already been discarded. What survived from Ockham was the imperative. What did not survive, and had to be reinvented, was the reason to obey it.

That reinvention is easy to miss precisely because the vocabulary never changed. A working scientist reaching for the razor today says exactly what a fourteenth-century Franciscan said, prefer the account with fewer things in it, and the continuity of the sentence disguises the discontinuity underneath it. The Latin tag Thorburn went looking for and failed to find was, in a sense, invented twice: once by later readers who needed a slogan to summarise Ockham's theological arguments, and again by a scientific culture that needed a slogan to summarise its own unrelated preference for economical models. Both borrowed his name for a principle he would not straightforwardly have recognised as his.