Vox Day is a Christian blogger and author of The Irrational Atheist. We have agreed to a friendly dialogue about the reasons for our beliefs, though we’ll try to avoid regurgitating all the usual arguments for and against the existence of God. See our previous letters.
I’m glad you’re working with me to establish some criteria for “good explanations” before we start proposing good explanations for S:
(S) Humans often take pleasure in the involuntary and undeserved suffering of others.
It amuses me that even though we are now discussing philosophical technicalities for a bit, our readers still manage to find a way to hurl insults at whichever one of us they see as the “opponent.”
I suspect your training in philosophy or economics helps you to see the importance of our “philosophical spade work.” For example, to determine whether or not the United States is in a recession, it matters very much how we define the term “recession.” Likewise, to determine what is a good explanation of S, it matters very much how we define the term “good explanation.”
I said that one model of explanation says that x is a good explanation of y if x is a potential explanation of y, and if:
x possesses the following explanatory virtues to a greater degree than any other known potential explanations of y: testability, consistency with background knowledge, past explanatory success, simplicity, ontological economy, informativeness, predictive novelty, explanatory scope, and explanatory power.
You replied that “some of the virtues cause you to artificially limit the investigation of the unknown.” In particular, you think that testability, consistency with background knowledge, fitting within a tradition of past success, simplicity, and ontological economy are not useful criteria for finding a good explanation.
You do, however, see value in explanatory power, informativeness, predictive novelty, and explanatory scope. Like me, you would give explanatory power less priority than the other three.
I don’t need to keep all of my proposed explanatory virtues. Nobody has an ‘authoritative list’ of explanatory virtues. But allow me explain why I think the explanatory virtues you reject are more valuable than you seem to think. Then perhaps you can either come to agree with me or explain why you think I am wrong.
I wrote earlier that philosophers themselves are divided over how to give a good account of explanation, and many are pessimistic that a good account can be given. Neither of us has the time to offer a thorough defense of explanationism or any other account of explanation. Even those who take an explanationist approach will argue over which explanatory virtues to use – or as philosophers say, they will argue over the desiderata. But hopefully we can settle on something that will work well enough for our purposes here, since the main reason you are a Christian is that you think Christianity offers a good explanation of evil, and one of the main reasons I am an atheist is that I think Christianity offers a terrible explanation of evil. (We have chosen to focus our debate over which worldview offers the best explanation of a particular kind of evil: the fact that “Humans often take pleasure in the involuntary and undeserved suffering of others.”)
My list of explanatory virtues was not meant to be a list of necessary or sufficient conditions, but rather a list of highly desirable conditions. What I mean is that a good explanation need not possess all the explanatory virtues, but the more virtues it has in greater degree, the more likely it is to be the best explanation.
Here, then, are your rejected criteria:
The problem with [your] definition by explanatory virtue is that some of the virtues cause you to artificially limit the investigation of the unknown by handcuffing it to the parameters of that which is presumed to be known. This all but guarantees systematic errors based on incorrect assumptions of the past. While it would certainly be ideal for a good explanatory hypothesis to be testable, but that is simply not possible in all cases. It therefore sets an artificial and incorrect technological limit on the process; for example, the x-ray hypothesis was correct regardless of whether it was possible to test for them.
Consistency with background knowledge is irrelevant. Fitting within a tradition of past success is similarly irrelevant. Simplicity is irrelevant too. This is philosophy, not interior design. Ontological economy begs the question of what is “necessary”; Occam’s Razor is a shortcut, not a reliable rule. On the other hand, informativeness is correct, predictive novelty is both applicable and useful, and explanatory scope and power are reasonable. I would give priority to informativeness, explanatory scope, and predictive novelty.
You say that a criterion like testability can “artifically limit” our ability to offer certain explanations. But of course. This is the entire point of having criteria. Without criteria, any old theory could be a ‘best explanation.’ Without criteria, we might as well explain gravitation with reference to “invisible gremlins living in every quark and lepton, and their affection for each other.”
The problem with the Gremlin Theory of gravity illustrates our need for the criteria I suggested – criteria that have been extremely successful in producing probable explanations that genuinely increase our knowledge in ways that allow us to create airplanes and vaccines and atom bombs.
Now, I must confess I’m puzzled by your choice of criteria to accept or reject. For example, you accept informativeness and reject testability. But to most philosophers, the whole point of informativeness is to allow for testability! As I defined it earlier, informativeness refers to a theory’s capacity to allow us to deduce the precise details of its effects. But what use are specific predictions if we cannot test them against reality?
Another puzzler is your acceptance of predictive novelty, given your rejection of testability. Predictive novelty refers to a theory’s capacity to predict not just known facts, but previously unknown facts. But again: what good is predictive novelty if we cannot test these novel predictions against reality?
You also say consistency with background knowledge is irrelevant, but I don’t see why. Surely we want a theory that generally fits with what we already know. One problem with the Gremlin Theory of gravity is that it doesn’t fit with many things we do know – for example that thought and attraction seem to depend on complex neuronal processes. Or, another example: One problem with a theory that mammals evolved before fish is that it doesn’t fit with millions and millions of pieces of evidence in the fossil and genetic records.
I also think that fitting within a tradition of past success is a useful criterion, but since I think it is one of the least useful, I will simply drop it for the sake of space in our conversation.
Almost as shocking as your rejection of testability as a criterion for explanation selection is your rejection of both simplicity and ontological economy as useful criteria. For most theistic philosophers engaged in offering “God did it” as the best explanation for certain features of this world, simplicity and ontological economy are considered preeminent among criteria for explanation selection. See, for example, Richard Swinburne’s Simplicity as Evidence of Truth.
Because you shockingly reject what most explanationist philosophers would consider to be the most important explanatory virtues, I suspect I may be misunderstanding you. I look forward to your clarifications. I am hoping we can find enough common ground to move forward, though perhaps here is where we diverge. If we diverge on this more fundamental point, it will do us little use to argue over points for which our divergence can be explained by this higher-order disagreement.
Previous post: News Bits