A good case against God

[CliveStaples]

Certianly there are threads on the internet where theists can present evidence for the existence of God. This is not one of them. This is the atheist's chance to do likewise, and I have tried to defend with reason that even if there is no demonstrable evidence for God it does not justify atheism. Go ahead and be agnostic if you want, but if you refuse to give any argument against God or belief in God that can stand up to scrutiny, I contend that your atheism is unjustified.

So really in this thread I am not arguing that God exists.

I am arguing that the absence of demonstrable evidence for God doesn't justify atheism.

I am also asking atheists to give good reasons to think there is no God, and I can't say that I think I have seen any.

And that one shouldn't rule out God simply because of the apparent absence of evidence.

If there are any agnostics here I would invite you to join me.

Well, I'm not an atheist, but I can think of a reason that absence of evidence for God justifies (weak) atheism.

First, methods of drawing inference rely on evidence. There is a difference between absence of evidence and evidence of absence (looking in a cage and seeing empty air is evidence of the absence of a polar bear; not looking into the cage at all is absence of evidence), and in the absence of evidence, we cannot draw valid inferences.

For suppose we could. Suppose we were to assume a principle whereby a hypothesis would be answered in the affirmative (or in the negative) in the absence of evidence, with confidence 0 < c < 1. In order to be consistent, c must be constant across all hypothesis tests (if we ask the same question again and again, each time with no evidence, we shouldn't become more or less confident in our answer).

So suppose we are asking whether some hypothesis H is true, but we have an absence of evidence. We then affirm H with confidence equal to c. Now, let us ask another question--whether ~H is true. We will also assign this c.

However, P(H) = 1 - P(~H), since P(H or ~H) = 1, and H and ~H are mutually exclusive, hence P(H or ~H) = P(H) + P(~H) = 1.

Thus c = 1 - c, hence c = 0.5

Now, any argument used to justify believing H on the basis of our confidence that H is true applies equally to ~H. Thus there is no way to probabilistically distinguish them; we must either arbitrarily draw an inference (which negates the whole point of having a method of inference), or decline to draw an inference.




This argument could be further developed using utility theory; choosing between H and ~H could be done based on both the confidence we have in them and the expected utility of each.