Vous êtes sur la page 1sur 5

GOD AND REV.

BAYES

VICTOR J. STENGER

For Reality Check in June, 2007 Skeptical Briefs.

Draft of Monday, April 16, 2007 12:20 PM for comments only. Do not quote, copy or

distribute.

Bayes' theorem (Thomas Bayes, d. 1761) provides a means for directly calculating the

probability for a statement being true based on the available evidence. In a 2003 book

The Probability of God (New York: Three Rivers Press), Stephen Unwin attempted to

calculate the probability that God exists. Unwin's result: 67 %. Physicist Larry Ford

(private communication) has examined Unwin's calculation and made his own estimate

using the same formula. Ford's result: 10-17. In what follows I present Ford's nicely

concise analysis, slightly modified.

Here's how the Bayesian method works. Let P(G) be the prior probability that a

proposition G is true. Now, suppose we have some new evidence E. Let P(G|E) be the

probability that G is true in light of the evidence E. Let P(E|G) be the probability that E is

true if G is true and P(E|G*) be the probability that E is true if G is false. Then is it easy

to prove that

P(G)P(E | G)
P(G | E) = (1)
P(G)P(E | G) + ⎡⎣1 − P(G) ⎤⎦ P(E | G*)

This is Bayes' Theorem.

Let G be the proposition that God exists. Unwin rewrites (1) as


Pbefore D
Pafter = (2)
Pbefore D + 1 − Pbefore

where

P(E | G)
D= (3)
P(E | G*)

he calls the "divine indicator," which represents how more likely the evidence E would be

if God exists compared to him not existing.

Unwin then puts in some numbers. He takes the prior probability of God existing,

that is, the probability before any evidence is submitted, to be Pbefore = 0.5. Then he

introduces a series of six observations and estimates the divine indicator D for each. At

each step he calculates a Pafter and equates that to Pbefore for the following step.

1) The evidence for goodness, such as altruism: D = 10 ⇒ Pafter = 0.91.

2) The evidence for moral evil, done by humans: D = 0.5 ⇒ Pafter = 0.83.

3) The evidence for natural evil (natural disasters): D = 0.1 ⇒ Pafter = 0.33.

4) The evidence for "intra-natural" miracles (successful prayers, etc.): D = 2 ⇒ Pafter =

0.5.

5) The evidence for "extra-natural" miracles (direct intervention by God in nature): D = 1

⇒ Pafter = 0.5.

6) The evidence for religious experience (feeling of awe, etc.): D = 2 ⇒ Pafter = 0.67.

Unwin then adds a boost based on faith raising the final probability of God to

0.95.

Now let's look at Ford's alternate estimate of these numbers. First he notes,

"propositions that postulate existence have a far less than 50% chance of being correct."
Another way to say this is that absence of any evidence or other reason for us to believe

some entity such as Bigfoot or the Loch Ness monster exists, it is highly unlikely that it

does. So the prior probability of God should be more like one in a million or less. So let's

take Pbefore = 10-6.

With respect to the divine indicator, D, we must evaluate it for each kind of

evidence. Taking miracles for example, P(E|G) is the probability of the observed

evidence of miracles given God exists. We see no evidence of miracles and since God

should be producing them if he existed, this probability is small. On the other hand the

absence of evidence for miracles is just what we expect if there is no God, so P(E|G*) is

near one. Consequently, the divine indicator based on the absence of evidence for

miracles is D << 1. Let us go though the various indicators.

Unwin exhibits the typical theistic fallacy that goodness can only come from God

and assigns a high divine indicator D = 10 for this. Ford points out that we should see a

lot more goodness in the world than we do see, if God exists. So he assumes D = 0.1.

Ford notes that the existence of both moral and natural evil in the world is

evidence against Gods existence. Unwin seems to agree by assigning D-values less than

one, but not sufficiently low to describe the true situation in which millions die or suffer

needlessly each year from the evils of both humanity and nature. Ford's values of D =

0.01 and D = 0.001 for moral and natural evil respectively are far more reasonable.

Unwin thinks that miracles such as prayers being answered have been observed

and so assigns a diving indicator D = 2 to what he calls intra-natural miracles. However,

the scientific fact is that the best, controlled experiments on intercessory prayer show no
positive effects. This scientific results makes Ford's estimate of D = 0.01 in better

agreement with the data.

Unwin assigns D = 1 for extra-natural miracles where God intervenes directly in

nature. Since there is not a scintilla of evidence that God does this, including the fact that

no miracle was required to bring the universe into existence, Ford's estimate of D = 0.1

for this property strikes me as far too generous.

Finally, there is no evidence that so-called religious experiences have any divine

content. If they did, we would expect the people having them to return with information

about reality that they could not have known before the experience. These "prophecies"

could be tested scientifically to see of they came true. None ever have. So, instead of

Unwin's D = 2, Ford's D = 0.01 is also more reasonable.

In any case, here is the summary of Ford's calculation:

Pbefore = 10-6.

1) The evidence for goodness, such as altruism: D = 0.1 ⇒ Pafter = 10-7.

2) The evidence for moral evil, done by humans: D = 0.01 ⇒ Pafter = 10-9.

3) The evidence for natural evil (natural disasters): D = 0.001 ⇒ Pafter = 10-12.

4) The evidence for "intra-natural" miracles (successful prayers, etc.): D = 0.01 ⇒ Pafter =

10-14.

5) The evidence for "extra-natural" miracles (direct intervention by God in nature): D =

0.1 ⇒ Pafter = 10-15.

6) The evidence for religious experience (feeling of awe, etc.): D = 0.01 ⇒ Pafter = 10-17.

Of course, many of you are likely to say this is a silly exercise, that the numbers

used are a matter of taste and obvious prejudice. However, I think it is useful to go
through it anyway. The mathematically challenged are often awed by any sort of

quantitative calculation, which they are unable to evaluate, and are likely to view Unwin's

work as providing scientific support for their beliefs. It does no such thing. Unwin loses.

If anything his method demonstrates the high unlikelihood of God's existence.

Vic Stenger's latest book, God: The Failed Hypothesis. How Science Shows that God

Does Not Exist has made an appearance on the New York Times best seller list. His

website is at http://www.colorado.edu/philosophy/vstenger.

Vous aimerez peut-être aussi