The power of prediction

During a debate with Christian apologist William Lane-Craig, philosopher Alex Rosenberg made the assertion that theism does not have any predictive power. Asked for his response, Craig beat around the bush before admitting predictive power was a problem but focused on the explanatory power of theism in explaining the universe, something he claimed atheism was unable to do. The problem with such an answer is that prediction is at the very heart of explanation. By admitting that theism had difficulty making solid predictions about the universe that could falsify it as an explanation for reality, Craig conceded the point that theism's explanatory power was deficient, despite his attempt to prove the opposite.

When asked about the best explanation for the presence and qualities of the universe, people cite their worldview, most often theism of some flavour. To them, it is what makes the most sense. In more rigorous and technical terms,  explanatory power can be thought of as the degree to which the evidence fits the hypothesis in question. The problem comes when people of all different types of worldviews assert that theirs has the greatest explanatory power, as both Rosenberg (an atheist) and Craig (an evangelical Christian) argued. They cannot all be right: indeed, most or even all must be wrong. Clearly two people like Craig and Rosenberg can be sincerely convinced, so who (if only one) is misinformed, uninformed, disingenuous, or lacks understanding? There must be a way to find out who if anyone has the closest approximation to the truth, a way which can be accessed and tested by people of different religions and cultures.

Prediction boils down to the question: what would prove you wrong? If you are confident that your explanation is the right one, the best way to put it all on the line is to make a specific, testable prediction which your pet theory would fail if it's wrong. People often think that prediction is most useful for proving something right, but it doesn't work that way in real life or in science. It is simply far more useful to disprove a hypothesis than prove it true, because a thousand positive data points can be outweighed by one data point which the hypothesis cannot explain. If one prediction fails, then you need to go back to the drawing board to modify your explanation.

If efforts to accommodate the hypothesis to the evidence are to be successful, they must not be ad hoc, or simply made up to save an idea from falsification. If an accommodation is to rise above rationalization, it must lead to a testable prediction which could falsify the hypothesis. Different creationists currently use two mutually incompatible arguments for theism with regards to our place in the universe. Some point to the vastness of the cosmos and the tiny part of it where we have conditions suitable for our existence, and say "that proves God meant for us to live." Other creationists claim that the physical properties of the universe are perfectly fine-tuned by god for life. These competing claims cannot both point toward the existence of a designer. These rationalizations save a cherished belief from falsification, but for the same reason are ad hoc. When a hypothesis can explain anything at all, it really explains nothing.

Adding accommodations to an idea also increases its complexity, which violates the measure of simplicity, referred to as Ockham's Razor. Given two explanations with equal fit with the evidence, the simpler of the two is more probable.  To illustrate, many young-earth creationists use the Omphalos hypothesis, which insists that god created the universe only a few thousand years ago, but with apparent old age to look as it does now. It is possible, and would explain all the evidence, but adds so much complexity that it is irrational to accept. Furthermore, it is in principle unfalsifiable. To put it in technical terms, this example adds complexity for no gain in explanatory power.

Some accommodations are worth taking seriously because they make testable predictions to explain new evidence and help us focus research to more efficiently pursue greater understanding. When irregularities in the orbit of the planet Uranus could not be explained by Newton's law of gravitation, an eighth planet was predicted. The prediction was carefully calculated and fairly specific, so it was worth taking seriously. Sure enough, the planet Neptune was observed, so Newton's law was redeemed as an explanation. Furthermore, its tested prediction made the power of Newton's law to explain the planets' orbits much greater.

The problem is that theists rarely make predictions this specific and potentially damaging to their hypothesis. This could come from avoidance of the real risks associated with prediction, but could also stem from the very vagueness and ambiguity of the hypothesis itself. If the breadth of current and historical theist explanations for the universe shows us anything, it's that the use of "god" is infinitely flexible to accommodate any scientific or pre-scientific conclusion. That, however, decreases its explanatory power, not the reverse. If a hypothesis can be accommodated to all possible sets of data, then there's no way to tell if it's wrong. The difference is what kind of specific, defeasible predictions can be made which would invalidate the hypothesis, particularly difficult for one with so many different iterations. The number of different versions, many of which are contradictory, raises the question of whether the "answer" is far more unknown and problematic than the data it seeks to explain, violating the principle of economy (Occam's razor).

When asked what new evidence would falsify their cherished believes, many people answer like Ken Ham (in his recent debate with Bill Nye) and say "nothing." In one study, over 60% of Americans who were questioned stated that they would ignore scientific findings which disproved a cherished belief of theirs. They are set in their beliefs, which are virtually guaranteed to be wrong because they have severed the connection between evidence and explanation, despite all claims to the contrary. They have disconnected their beliefs from reality.

Some answer in the affirmative like Craig (on his website) and give such an impossibly difficult or high standard that they are guaranteed never to be proven wrong regardless of the truth of their claims. This is merely paying lip service to the importance of prediction while doing the opposite: making an after-the-fact accommodation with no testable way to falsify the explanation. The difference between the two answers is not very large, especially when compared with the myriad and specific predictions made by the various scientific hypotheses and theories which creationists attempt to replace with their particular religious belief. The end result is the same: a cherished belief held under a rock, safe from the blinding light of reality.