Skip to contentWillowOne's PositionsTagsWhat’s newMediaStatisticsSearch
WillowOne's Positions

Cumulative Case Argument

Form

The modern apologetic strategy, set out by Basil Mitchell (1973) and given a Bayesian form by Richard Swinburne (The Existence of God, 2004):

  1. No single argument for God is a proof, and the apologist does not claim otherwise.
  2. But each argument (cosmological, fine-tuning, moral, consciousness, religious experience, miracles) raises the probability of theism a little: each is a piece of evidence that is more likely if God exists than if he does not.
  3. Combined, the pieces make theism more probable than not. Swinburne puts the probability above a half, with religious experience as the strand that tips it.
  4. The case is like a rope of many strands, or a legal case built from circumstantial evidence: no strand bears the weight alone, and the whole holds.

My Positions

  1. A rope of broken strands holds nothing. A cumulative case requires that each piece be some evidence. An argument with a false premise has an evidential weight of zero, not "a little", and ten of them sum to zero. The strategy quietly converts "this argument fails" into "this argument is weak but positive", and then adds up the conversions. See the first general observation under Religious apologetics: ten leaky buckets hold no more water than one.
  2. The Bayesian version needs numbers it does not have, and leaves out the evidence against. Swinburne's probabilities are assigned by intuition. The same method, applied honestly, must also count the Problem of evil, Divine Hiddenness, the total dependence of mind on brain, the failure of prayer under test, the religious diversity of the world, and the natural history of religious belief, each of which is more likely if there is no God than if there is. Paul Draper's evidential argument from pain and pleasure does exactly this sum and gets the opposite result. The cumulative case counts the hits and not the misses (Cherry-picking, Texas Sharpshooter).
  3. The strands pull in different directions. The cosmological arguments deliver a cause, the design arguments a tuner, the moral argument a lawgiver, the ontological argument a maximal being, the resurrection a first-century Jewish miracle. These are not strands of one rope but ropes tied to different posts, and several of the posts are incompatible: the changeless cause of the Kalam Cosmological Argument cannot be the agent who acts in history, and the simple being of the Argument from Divine Simplicity cannot be a Trinity. The apologist assumes they converge because the same word is used at the end of each.
  4. Legal cumulative cases work because each piece is a verified fact. The fingerprint, the alibi that failed, the purchase of the weapon: each is established, and the inference is from facts to a hypothesis. In the theistic case each "piece" is itself a contested inference whose premises are the things in dispute. A pile of disputed inferences is not circumstantial evidence; it is the dispute, restated.
  5. In practice it is the Gish gallop in slow motion. The critic who answers the cosmological argument is told that it was only one strand; who answers the next is told the same; and the case is never where the critic is standing. An argument that retreats to wherever it is not being examined is not being defended.
  6. I have looked at each strand. They are set out in this folder, one by one, in the strongest forms I could find. None moves the needle. Honest arithmetic on a column of zeros gives zero, and I decline to be told that the column is long.

Sources

  • Mitchell, B., The Justification of Religious Belief (1973)
  • Swinburne, R., The Existence of God (2nd ed., 2004), ch. 14
  • Craig, W. L. and Moreland, J. P. (eds), The Blackwell Companion to Natural Theology (2009), introduction
  • Draper, P., "Pain and Pleasure: An Evidential Problem for Theists", Noûs 23 (1989)
  • Mackie, J. L., The Miracle of Theism (1982)
  • Oppy, G., Arguing About Gods (2006)

Also see