A rule of simplicity -- for example, Ockham's Razor -- is normative. In its traditional formulation, Ockham's Razor can be expressed as follows: one should avoid the needless multiplication of entities. What counts as needless -- that is, one's standard of simplicity -- does not matter here. What does matter is that there be a rational ground for deciding among hypotheses. This decidability presupposes boundary conditions among which is one or another standard of simplicity. Without some standard of simplicity, one cannot succeed in falsifying a hypothesis, because one cannot reasonably rule out ad hoc complications introduced to accommodate facts that would otherwise falsify the hypothesis.
A rule of simplicity expresses a relationship between a purpose sought in attempting to explain and a means necessary for achieving that purpose. A rule of simplicity is thus a norm expressible in the form of a conditional: if one wishes to achieve any purpose by an attempt at explanation, then one may not arbitrarily introduce complications.
Inasmuch as it is a rule, a rule of simplicity is not a statement of fact, although it is related to the facts which make up the process of scientific inquiry itself. A rule of simplicity involves a claim about a relationship between two aspects of that process -- namely, a purpose of the process and one of the conditions necessary to achieve it.
A determinist can admit a difference between conditional statements and conditional norms which he can explain on his own ground. Conditional norms might be explained, for instance, as expressions of emotion or as causally determined exhortations. The general rule of efficiency we have described might be explained, for instance, as a key component in the survival mechanism of the organism.
For the sake of argument we will grant any such deterministic account of normativity in terms of antecedent determining conditions. But in conceding this point, we mean to hold determinists to all the implications of determinism, including those implications which determinists ignore when involved in arguments with someone who disagrees with them. In particular, we wish to call into question the consistency between any determinist's account of normativity and his assertion of his position, involving -- as it necessarily does -- a rule of simplicity as an essential ingredient.
Any determinist hypothesis must be able to account for the existence in the world of conflicting attempts to account for the data of human experience -- there are positions that contradict determinism. A determinist might try to account for this fact by saying that both positions are determined effects of different sets of antecedent conditions.
Nevertheless, every determinist makes the claim that his account of the data is superior to his opponent's, and therefore ought to be accepted in preference to the alternative position. The question is, what meaning can a determinist attach to the word "ought" in this context? Certainly no determinist can mean what anyone who would disagree with him would mean by saying that we ought not accept determinism. Someone rejecting determinism can distinguish between the force of a norm and the force of determining conditions. But, any determinist must say that among the sets of determining conditions there is one set of determining conditions that determines him to say "ought" and determines whatever effects follow from his utterance of "ought." And he must give the same account of his opponent's utterance of "ought." This result will not seem odd to a determinist; it follows logically from any form of the determinist hypothesis.
On his own account of "ought," then, a determinist is perfectly able to say we ought to accept his position and ought not hold the contradictory position. But on those same grounds he must also grant that someone who has articulated a contradictory position is equally able to say that we ought to accept his position and ought not accept a determinist hypothesis.
No determinist can avail himself of a distinction between positions in fact maintained and positions justifiably maintained in any sense of that distinction which a determinist account would preclude. Where normativity is explained in terms of antecedent determining conditions, the exclusion of any position can be achieved only be excluding the very articulation of that position. But inasmuch as determinism is a more economical account of a set of facts that initially present themselves as including the naively realistic interpretation of the experience of choice -- which any determinist hypothesis explains as an illusion -- the contradictory position is necessarily articulated whenever any determinist position is articulated.
It follows that a determinist hypothesis cannot exclude its contradictory in the only sense of "exclude" that is available to a determinist. Any determinist hypothesis implies the impossibility of excluding its counterpositions, but necessarily presents its own counterposition in its very articulation. But a determinist, in arguing with his opponent, precisely does want to exclude the contradictory position. Otherwise there would be no point in the determinist's entering the argument, because the utterance of a sentence without the intention of excluding the contradictory is not a statement.
Joseph M. Boyle, Jr., Germain Grisez, and Olaf Tollefsen
"Determinism, Freedom, and Self-Referential Arguments"
The Review of Metaphysics 26 (1972-3)