Professor John McCarthy
Father of AI

Scientific Forms of the Religious Hypothesis

6 April 2008

When Laplace showed Napoleon his treatise on celestial mechanics, Napoleon asked him what place God had in his theory. Laplace replied that he had no need for that hypothesis. We also have no need for it, because science has been successful, and science is the best approach to solving the mysteries that remain. Let's consider religion anyway.

The religious hypothesis in its narrowest form is that the universe was designed and created by an intelligence. Another form is that some superhuman power frequently intervenes in life on earth. These hypotheses are distinct; one could be true without the other, both could be true or neither. One reason for considering these hypotheses is that when people understand computer programming well enough, people will be inclined to program evolutionary processes in a computer that might develop intelligence and might be inclined to intervene in the ``worlds'' they have created.

We can therefore consider the hypothesis that our world is a computer program or some other designed system. There are two interesting questions:

1. What would constitute evidence for creation or intervention?

2. Is there any such evidence? I don't see any.

3. Is it worthwhile to look harder for such evidence? My answer is that it is moderately worthwhile, although I'm not sure I would favorably review a proposal to the National Science Foundation for research aimed at it. There would have to be a good idea about where to look. Probably it's best to regard thinking about it as a leisure time activity for people with inclinations to speculation.

Note that this form of the religious hypothesis may not satisfy anyone's ``spiritual needs'' nor is it necessarily compatible with existing religions. It also won't console people for the death of their children or offer compensation in an afterlife for suffering and injustice in this life. It may be that the decline of religion in prosperous countries in this century is partly a consequence of prosperity and democracy having reduced the need for consolation as well as having popularized scientific views.

Indeed my speculations about religious hypotheses will not attribute to the hypothesized ``intelligence'' any special interest in human affairs, any sympathy with humanity, or any desire to be known or worshipped, or any tendency to hear prayers. If I were to write, run and observe a program that might evolve intelligence, I don't think I would be inclined to want it to print out ``John McCarthy is great''.

Since the advent of science, many religions have become ``purified'' and lost their previous scientific goals and pretensions. Indeed there is a tendency to deny that they ever had such pretensions. Nevertheless, before science existed, there were religious hypotheses about questions now treated by science. Here are some examples:

1. Thunder and lightning are caused by Thor throwing his hammer.

2.

God gave Noah that rainbow sign
That it won't be water but fire next time.

This explains rainbows by ascribing a purpose to them.

3. The reason there are ants is to persuade people not to be lazy.

4. Churches are often damaged by lightning, because the evil spirits of the air hate the houses of God. The alternate explanation is that churches are often struck by lightning by God in order to show the congregations that God disapproves of certain recent sinful behavior. (See A. D. White's ``The warfare of science and theology in Christendom''). Persistence of these explanations delayed by 100 years the adoption by some churches of Benjamin Franklin's 1757 invention of the lightning rod.

This kind of explanation of natural phenomena has been superseded by scientific explanation. The latter makes more satisfactory theories, and also leads to technology that works, while the technologies based on religious hypotheses, such as prayer, sacrifices, astrology and sorcery, don't work.

Pre-modern religion had both scientiic and technological aspects. Specific events, like sinkings of ships or victory in war were ascribed to the actions of one or more gods. That's the science. The technology consisted of actions aimed at obtaining the favor of God or gods. These included sacrifices and prayer. When a bad outcome occurred anyway, there were always excuses - not enough prayer or the actions of unbelievers.

It is interesting that the Greeks and Jews both mentioned human sacrifice in their past, but by the time they were literate, neither people did it. The Greeks and Romans both accused the Carthaginians of sacrificing babies. One can imagine the Carthaginian priests saying, as the Romans besieged Carthage, that the reason for the defeat was that not enough babies had been sacrificed.

Excuses

When an action doesn't have the predicted effect, excuses can always be found, and sometimes they are correct. However, it is common, and maybe usual, that the adherents of a theory keep making excuses for its failures long after they should give it up. Failure of prayer and animal and human sacrifice often met the excuse that the prayer wasn't intense enough.

Francis Galton (1872) asked whether the numerous prayers for the health of the British royal family gave them longer lives. They didn't.

It's not only religious theories for whose failures excuses are made. Medicine provides many examples, e.g. for the practice of bleeding the patient to relieve fever (said to have killed George Washington.) In medicine and in some other activities, the important way of not falling for excuses is the double blind test comparing a treatment with a placebo. Neither the subject nor the experimenter knows whether the subject has received the treatment or a placebo.

While the above theological explanations of natural phenomena are inferior to non-theological explanations, this can be ascribed to the naivete of pre-scientific cultures and to their self-absorption. We could try for theological explanations that are less parochial. For example, we could try to explain some aspect of the theory of relativity as fulfilling some purpose of the God that designed the world in this way. Indeed Einstein and some other scientists look for beautiful theories, but none of them ascribe any but aesthetic motivations to God. Also it's hard to tell whether their references to God are just metaphorical.

As another example, note that the religious often cite the complexity of life as an argument for the implausibility that present day life is purely the result of random variation and natural selection. [The best explanation I have seen of how complex features arise by natural selection is given in Climbing Mount Improbable by Richard Dawkins.]

The alternative religious hypothesis, supposing that the geological evidence for the succession of species isn't a fake on God's part, must ascribe a purpose to the particular observed succession of species. The British biologist J. B. S. Haldane is supposed to have been asked by a bishop what his biological studies told him about God's purposes. Haldane is said to have replied that God seemed to have ``an inordinate fondness for beetles''. Indeed a theological explanation of the succession or distribution of species would have to account for the 300,000 species of beetles. [2003 note: Lacking a theological explanation, biologists have begun looking for a biological explanation of why there are so many species of beetle].

The clockmaker religious hypothesis doesn't have such obvious disadvantages. God created te world in the first place and set it running. A religious explanation would involve saying why this kind of world was chosen and not some other. When we know more physics, someone may explore such hypotheses.

Now let's explore the religious hypothesis from the other end. Suppose we were to program a world and set it running and it evolved intelligence. Could these intelligences determine or even plausibly conjecture that they were part of a computer program, and could they infer anything about us, the programmers?

Maybe! It depends on what kind of program we wrote.

March 1989

Suppose the prgram were very large and involved many detailed decisions. Suppose we made these decisions on the basis of some ideas about what kinds of events we preferred to happen in our simulated world. Imagine that the program had thousands of such decisions made by hundreds of thousands of programmers over hundreds of years. It might then turn out that the scientists within the simulated world could account for their world's features most economically by figuring out our purposes. They might even be able to identify the styles of the different programmers of the different aspects of the world and their characteristic mistakes. This would depend on the how the detailed decisions related to features observable by them.

Correspondingly, a present day intelligent design advocate might try for a theory of the Designer's decisions. Could he make a theory of the Designer's purposes that accounted for the number of species of different forms of life more economically than anything the biologists could devise? Alas, they aren't even trying to enumerate, let alone account for, the Designer's decisions. I think the "intelligent design" people are just kidding us.

The Polish science fiction writer Stanislaw Lem raised the question of the human programmer's moral obligations to his simulated creations. Might it be wrong to program a world of suffering or to terminate a world that had a civilization in it? I have to confess I have few moral intuitions about that, one way or the other. I suspect that concepts of morality get rather tenuous when taken much beyond human affairs.

By the way, I count myself as an atheist rather than an agnostic. My criterion for being an atheist is believing that the evidence on the god question is in a similar state to the evidence on the werewolf question.

Religion as an industry

We can count the number of ministers, priests and other employees of religious institutions and compute a fraction of the GDP that goes into religion. If tithing, i.e. contributing 10 percent of ones income were universally practiced, it would be 10 percent of the GDP. We could ask whether this fraction is increasing or decreasing and make comparisons among countries of regions within countries.

A not very determined look at census data suggest that half the American population belong to churches and give an average of 713 dollars per year. I suppose this would support about a million people. Better statistics are surely available.