In “Why Anything? Why This” (On What Matters, Vol. II, pp. 623-648), Derek Parfit offers an ingenious line of argument against the insistence that the existence of the Universe be given a non-coincidental explanation. While I am on the same page with Parfit about this point, I think there is some larger issue going unnoticed and that he should have pushed his argument further to a deeper level.
Let me start from a ostensibly minor error his makes when refuting the Axiarchic View, which he formulates generically, as consisting of the following three claims (p. 633):
(1) It would be best if reality were a certain way;
(2) Reality is that way;
(3) (1) explains (2).
Parfit gives (1) and (2) a pass, and instead focuses on scrutinizing (3):
(1) is an ordinary evaluative claim, like the claim that it would be better if there was less suffering. The Axiarchic View assumes, I believe rightly, that such claims can be in a strong sense true. (2) is an ordinary empirical or scientific claim, though of a sweeping kind. What is distinctive in this view is claim (3), according to which (1) explains (2).
While I also think (3) is problematic [note: for reasons slightly different from Parfit’s, to which I will return later], I am surprised that he lets (1) and (2) off the hook so easily, without arguing that the claim that “an ordinary evaluative claim [like (1)] can be in a strong sense true” and the claim that “(2) is an ordinary empirical or scientific claim” cannot be both true at the same time.
To begin with, notice that “it would be best if …” is a normative/evaluative statement [note: I prefer to use “normative” but since Parfit here uses “evaluative” I will leave it at that for a moment], and that reality being a certain way can be interpreted to mean either that reality has a certain normative/evaluative property, ν, or that it has a certain non-normative/non-evaluative property, φ. For example, “lack of gratuitous suffering” is a normative/evaluative property, to the extent that “gratuitous suffering” is a “thick concept,” whereas “obtainment of rectangular planets” is apparently a non-normative/non-evaluative property.
Accordingly, there are two ways to rewrite (1) and (2). On the one hand, we may have its normative/evaluative version:
(1A) It would be best if reality had a certain normative/evaluative property, ν;
(2A) Reality has that normative/evaluative property, ν.
Now, since (1A) essentially claims that if a normative/evaluative condition is satisfied, then a normative/ evaluative conclusion follows (e.g. “it would be best if reality were such that there were no gratuitous suffering”), it indeed “can be in a strong sense true” as it is something that is apt for justification through a priori reasoning. However, the correspondingly claim (2A) is not “an ordinary empirical or scientific claim” at all, as it makes a claim on a normative/evaluative, not an empirical, property of reality (e.g. “reality is such that there is no gratuitous suffering”).
On the other hand, suppose (1) and (2) are reformulated as:
(1B) It would be best if reality had a certain non-normative/non-evaluative property, φ;
(2B) Reality has that non-normative/non-evaluative property, φ.
Then (2B) is indeed “an ordinary empirical or scientific claim” (e.g. “reality is such that there are rectangular planets”), but to what extent is (1B) still “an ordinary evaluative claim” that “can be in a strong sense true”? The claim that it would be best if there were rectangular planets doesn’t make sense unless we supply an account of how the existence of rectangular planets (or likewise other non-normative/non-evaluative properties) actuates in reality a certain normative/evaluative property, ν, the possession of which in turn makes reality normatively/evaluatively superlative.
To be sure, moral naturalists have long claimed that normative properties are reducible to, or supervene on, non-normative “natural” properties. To illustrate, suppose you believe that the badness of suffering per se can be explained away by purely natural facts (e.g. “suffering is no more than the experiencing of pain, and pain is bad, simpliciter”), then you could regard the claim that it would be best if there was least suffering as an instance of (1B) rather than that of (1A). On the face of it this claim sounds “in a strong sense truth.” But just think about why the Logical Problem of Evil has been replaced by the Evidential Problem of Evil: if suffering is a non-normative fact then what really plays the role in normative argumentation becomes “gratuitous suffering” rather than suffering per se. Therefore, instead of “it would be best if there was least suffering” what you would have to prove is “it would be best if there was least gratuitous suffering.”
Now you could either acknowledge that the claim that it would be best if there was least gratuitous suffering is an instance of (1A) instead of (1B), or (as a staunch moral naturalist) keep on trying to reduce the gratuitousness of suffering to a set of non-normative/non-evaluative facts, so as to keep the claim as an instance of (1B). After all, this is what moral naturalists do. – The problem is, however, that Parfit is himself a moral non-naturalist (so am I), which means he shouldn’t have accepted (1B) without questioning from the start.
In a nutshell, the lumping together of (1) and (2) is misleading (at least for non-naturalists such as Parfit and myself) [note: hopefully I could later return to the naturalism-vs.-non-naturalism debate & the compatibility issue between naturalism and the Axiarchic View], and obscures one of the intractable normative/evaluative assumptions underlying the Axiarchic View.
What, then, does this have to do with (3), and with the “larger issue going unnoticed” I have alleged? The next post will discuss.