Stardate 20030506.1827

(Captain's log): It's time to get all metaphysical. In this post I made a comment in passing that I was certain that atheism could not be proved. As soon as I posted it, I knew I'd have to back that up and started thinking about it.

"Atheism" isn't a unitary whole; it's actually a collection defined by a negative, consisting of all people of all beliefs who do not think that there are any deities. But that leaves a lot of room for beliefs in other kinds of supernatural forces or non-material influences, and because of that "atheists" are all over the map in terms of what they believe. My own form of atheism is known as "mechanism" or "materialism" and refers to the stronger statement not just that there are no deities, but that there are also no other supernatural forces. A mechanist believes that the material universe is all that exists, and that everything we see around us is a manifestation of matter and the way it interacts according to the laws of physics.

To many people a mechanist is the paradigmatic atheist, but there are misconceptions about us. I've run into Christians who thought that atheists were people who knew that God existed and knew that Jesus was the Christ but refused to accept Jesus as savior for some unfathomable reason. For such people it was inconceivable that someone might think that there was no God at all and that Jesus, if he lived, was just a man. (As were Mohammed and Siddhartha.) For them it was a priori true that God exists, and extremely difficult for them to accept the idea that someone else might not agree with that.

Unfortunately, there's another problem with how atheism is perceived, and in this case it's the atheists' own fault. To take a slight turn to the side for a moment, I've noticed that in the case of computer religion that there are some people who cleave to minority operating systems such as OS/2 or MacOS or Linux because they truly think they're good and enjoy using them, but that there are others who tend to adopt any one of those mostly because they're not Windows. Such people are not actually fans of OS/2 or MacOS or Linux as such and may not even be very knowledgeable about them, but tend to be vocally anti-Microsoft in any kind of advocacy discussion and often are much more militant about their claims of superiority of whichever alternative to Windows that they're using for rhetorical purposes. One can generally differentiate these two kinds of users in pretty short order based on what they say and how they say it.

By the same token, there are some people who are atheists because they come around to the idea of atheism as being the most acceptable answer, for whatever reason. I'm one of those. But there are a lot of people who become vocally atheist mostly because it isn't Christianity, and try to wield atheism as a rhetorical weapon in their crusade to convince Christians that Christianity itself is wrong. To that end, they will try to claim that atheism is better. Like all zealots, such advocates can become a pain in the ass; they're usually confrontational and intolerant and actively seek out strife as they engage in anti-Christian evangelism. Because the atheists like me generally don't seek out "the enemy", it's the zealots who are most visible and establish the reputation we all must live with.

This is most commonly a failing of the young, just as operating system zealotry is, and it's not at all uncommon for a young person to vocally embrace atheism, or some very strange religion such as "The Church of the Goddess", as part of a more general revolt against parental authority, which on some level all young people go through as part of the maturation process. It's the same kind of thing as dyeing their hair green or getting a piercing or tattoo; it's as much about scandalizing the grownups as anything else. Most people grow out of that, fortunately.

I certainly can't speculate about the specific motivation of The Raving Atheist, but on his site he has a page which outlines his manifesto in three parts. And like many evangelistic atheists, he seems to strongly equate "God" with "The Christian God", and to try to assume that the kinds of claims made about the Christian God are somehow universal to all theistic religions, even though this is not true. There are many religions where the deities are not omnipotent, not omniscient, and indeed may not even be immortal. (Norse mythology includes a story of how Balder is killed, for instance.)

It is the second of his three points with which I most strongly disagree, and it is short and I quote it in full:

Second, Atheism is not merely one possible theological theory among many. Rather, it is the only true, provable theory, and all other religious theories are false and delusional. The mere fact that you believe or have faith that god exists does not make it so, anymore than unicorns, ghosts, leprechauns would exist simply because you believed in them or had faith in them. Nor does the fact that you have a legal right to believe in god prove that god exists. Similarly, the fact that American law purports to be “neutral” as between theism and atheism does not mean that the theories are equally plausible.

For those who go beyond embrace of atheism for themselves and need it as a weapon with which to bludgeon Christians, this is essential. It's to his credit that he recognizes that and states it openly.

To say that atheism is both true and provable is a very strong claim, and it's made in a way which covers all theistic religions including those whose deities bear no resemblance to the Christian God. Note that he is not merely claiming that theistic religious theories are inferior to atheism, or less reliable; he's making the absolute claim that they are false, and that atheism is true and that it's possible to prove that it is true and to prove that all the others are false. But there isn't any way that mechanistic atheism can be directly proved. The only way to have any chance at all to prove it is via reductio ad absurdum.

In that form of proof, a given statement must be taken as true because it has been demonstrated that all alternatives to that statement are false. So I would restate the claim thusly:

It is impossible to construct a theory of the universe which includes a deity without either creating an internal contradiction or yielding a prediction which can be disproved empirically.

If that is true, then atheism is indeed proved. But proving this to be true is extremely difficult. For one thing, demonstrating that any single religion (e.g. Christianity) is either internally inconsistent or is empirically false isn't sufficient. Doing so for dozens of them isn't sufficient. It is necessary to prove that all existing religions are false, and that any theistic religion we can conceive of is false even if it's not actively practiced by anyone.

Reduction ad absurdum is a perilous form of proof, because you really do have to disprove every possible alternative statement, and sometimes people using this argument screw up and don't manage to even identify all the possible alternative statements, or actually prove that they actually are all false. (And it's been known to happen that some inconvenient alternatives will be deliberately ignored.) If the identification of the set of alternatives is incomplete, the argument isn't rigorous.

It's difficult to see how an advocate could actually demonstrate such a thing, but it's very straightforward to prove it wrong. All one must do is to show one single case of a religious theory which is consistent and which cannot be disproved empirically, and that can be done. I hinted at it in my post when I referred to "the apathetic God". If such a single counter example exists, then the proof of falsity of all alternatives to atheism fails, and atheism is not proved.

Update 20040611: I wanted to expand on that a bit, because there seems to have been confusion (possibly deliberate) about who bears burden of proof in certain cases.

Epistemologically speaking, burden of proof is always incumbent on the person who makes a claim, irrespective of the nature of that claim. But some claims are easier to prove than others. Weak claims have narrow scope, are more easily proved, and generally are not controversial.

Belief atheists like me make weak claims. I do not say that "there is no God". I say that "I believe that there is no God". The claim is about my belief. The only way my statement can be false is if it can be shown that I don't actually hold that belief, and since I'm ultimately the only one who really knows whether that's what I actually believe, it is very difficult for anyone else to disprove my claim. And if I really do hold that belief, my statement would be true even if it turned out there was a God.

The claim, "I believe in X" is a claim about the existence of a belief. If I actually do believe X, then that statement is true; I'm telling the truth about what I believe. If I don't actually believe X, the statement is false; I'm lying about what I believe. But that has nothing whatever to do with whether X itself is true or false. If I say, "I believe that pigs have wings", then if I really am deluded in that way, the statement would be true despite the fact that pigs do not have wings. (Yet. Wait until the genetic engineers get going.)

My statement that "I believe there are no gods" is a weak statement, and part of why it is uncontroversial is that even if my statement is a fact, it doesn't have logical consequences for anyone else's religious belief. It's a claim about me, not a claim about the universe.

It is true that I believe there is no God. My statement is a fact. But that fact doesn't disprove Christiantity; all it proves is that I am not a Christian.

If a Christian says, "I believe Jesus is the Son of God", then he speaks truly if he really does hold that belief. That is also a weak claim only about him. If the Christian really does hold that belief, an atheist must accept his statement as being true even though an atheist doesn't grant the existence of God. But an atheist also need not find that fact troubling. Even though it is a fact, it doesn't disprove atheism. All it proves is that the particular Christian isn't an atheist.

Some atheists like RA do nonetheless seem to find such facts troubling. Others like me do not.

If a Christian says, "It is a fact that Jesus is the Son of God" then he has made a strong claim, one which applies to the universe. If his statement is true, then atheism is wrong. But it is the Christian who is making the claim, and therefore the Christian assumes the burden of proof for that claim. If the claim is true it's a real problem for atheists, but atheists are entitled to ignore that Christian and his claim until and unless he actually does come up with unambiguous proof. We do not have to take his claim seriously until he satisfies the burden of proof.

And when a proof atheist says, "it is a fact that there are no Gods", then he too has made a strong claim about the universe. That is what the Raving Atheist does claim. And because he is the one making a claim, burden of proof is on him. If the claim is true, it's a problem for Christians. But until he satisfies the burden of proof, Christians may ignore him and go about their business.

He isn't permitted to say that Christians (or other theists, but it's Christians who are RA's target) must prove that God exists, and if they cannot then they have to grant that his strong claim is true. That's an example of a fallacy known as "shifting the burden of proof". He is the one who is making a claim, so he is the one who must provide proof if he wants his claim taken seriously.

The difficulty for proof atheists is that it's damned hard to prove a negative, especially one as expansive as this. The reductio ad absurdum is pretty much the only way to do it. If it can be shown that every conceivable religion based on every conceivable deity or pantheon of deities either will be required to claim something which can be demonstrated to be empirically false, or will be required to claim something which is logically impossible, then it means there cannot be any Gods. That's the burden on the proof atheist. If he wants his claim accepted as fact, that's what he has to demonstrate. But that's damned hard because he has to get them all. It has to be complete. If he only shows that some religions embody contradiction or empirical falsehood, it isn't enough. It isn't enough to prove that all existing religions are wrong. He also has to prove that any conceivable religion which does not now exist also is wrong.

On the other hand, if it can be shown that there is one hypothetical religion which does not lead to logical contradiction, and which cannot be demonstrated false empirically, then that is sufficient to demonstrate that the proof atheist cannot be complete. That means that the resort to reductio ad absurdum fails and does not prove the strong claim.

My contention is that the Theory of Fred as described below is such a counterexample. It is not possible to show that the Theory of Fred is false. Therefore I contend that I have proved that it is not possible to use reductio ad absurdum to prove the strong claim that there are no deities. And I don't know any other way that a proof atheist could prove that no Gods exist.

Unless proof atheists come up with some way to prove their strong claim, and then proceed to actually prove it that way, everyone else is entitled to ignore their strong claim.

The basic idea goes like this: God, whom I tend to refer to as Fred in the discussion of this theory, used His powers to cause the Big Bang. Since then, he's been watching the universe as it has developed because He wanted to see what would happen (because He didn't know). But Fred does not interfere in the universe, and Fred did not design the outcome. For instance, Fred did not try to manipulate the initial conditions so as to cause humans as a species to appear; it's just one of the many things He has observed while the experiment continues. Too, humans have no souls and when they die they're dead; they're just part of the universe which resulted from Fred's one action in setting the whole thing off. Fred does not listen to prayer; Fred does not interfere. Fred just watches, and He's just as surprised by what's happening as we are.

He is not part of the universe and is not subject to its laws, and is capable of watching it in a way which does not affect it, quantum mechanics notwithstanding. If it pleases you to do so, think of Fred as running the universe as a gigantic computer simulation, where He can see what's going on by getting printouts every once in a while or by watching some sort of massive graphics display. Or perhaps He's feeling the lumps underneath the curves of space induced by all mass, Braille-style.

In Fred's creation, everything happens strictly mechanistically; the only thing Fred ever did was to start it all in a way which Fred deliberately made indistinguishable from a natural equivalent. Fred has made the active decision to do nothing which would perturb the natural process of development of the universe, and Fred makes no mistakes.

By the way, this doesn't come out of whole cloth. It's more or less a modern statement of what was known about 200 years ago as "Deism", which at the time was the closest anyone who was intellectually honest could be to atheism before the advent of modern science. Deism used God as an explanation for the existence of the universe. It posited that God created the universe and then went somewhere else.

According to the Theory of Fred the apathetic God, the resulting universe would be completely identical to a totally mechanistic universe, and therefore there could be no conceivable piece of empirical evidence which would be different in Fred's creation relative to a mechanistic universe. There's no empirical way to tell them apart. Equally, there's no obvious intrinsic contradiction involved in the Theory of Fred. There is no case where the Theory predicts that a given statement is both true and false via different calculations.

The fact that we can't explain where Fred came from isn't a contradiction; it's just an unanswered and perhaps unanswerable question. The fact that we can't answer it doesn't mean the Theory of Fred is wrong. (And that argument is a dangerous one for an atheist, because there are lot of things we atheists cannot yet explain.)

Therefore, the Theory of Fred represents a counter example to my form of the strong statement regarding proof of atheism. There is no way for us to disprove the Theory of Fred the apathetic God, and reductio ad absurdum fails because we have demonstrated one alternative which cannot be proven to be false. (And we only need one.)

Within mechanistic atheism, you have people who think that atheism is somehow scientific and actually can be proved, and others who understand that atheism is a religious belief which is no more susceptible to actual proof than any other religious belief. I am among the latter group; The Raving Atheist clearly states his opinion that atheism is provable and stands with the former.

I cannot speak for him but I can speak of the general arguments made by those who do think that atheism can be proved, whom I'll refer to henceforth as "proof atheists", as opposed to "belief atheists" like me. I'm also going to be making reference to theists and unlike The Raving Atheist I'll be trying not to presume the universality of Christian dogma amongst all theists.

One argument that proof atheists raise regarding the Theory of Fred is that Fred as described isn't omnipotent or omniscient, and as such isn't really a God. But though Christianity ascribes those characteristics to its own deity, many other religions do not, and it is not logically required that all religions do so. It's very common in pantheistic religions for the deities to be less than totally omnipotent and totally omniscient; Zeus and Jupiter and Odin were not. And though I'm not an expert, I believe that none of the members of the Hindu pantheon are omnipotent and omniscient. For mechanistic atheism to fail, all that's necessary is to demonstrate existence of a deity who has the power to violate the laws of physics, and a lot of pantheistic gods have that degree of power even if they are not omnipotent. (Or they would have, if they actually existed.)

Mechanistic atheism denies the existence of non-omniscient and non-omnipotent deities, too. Thus the claim that Fred is neither omnipotent nor omniscient is uninteresting; Fred still represents a valid counter-example.

The proof atheist may try to claim that his opponent hasn't actually proved that the Theory of Fred is true. But the opponent doesn't need to. The reductio ad absurdum requires proof of falsity of all conceivable alternatives. Existence of any single alternative which can't be proved either true or false is sufficient to make the reductio ad absurdum fail. Burden of proof is on the atheist, not on the challenger.

The proof atheist will claim that the Theory of Fred as stated is in fact unfalsifiable, and then sometimes fumbles about a bit while trying to claim that this means that it is actually false. Of course, a statement which is unfalsifiable by definition can't be false, but we can be generous and adopt a more loose claim that it's sort of invalid because of its unfalsifiability.

It's true that it is unfalsifiable; it was deliberately constructed to be. But demonstrating unfalsifiability isn't disproof. It doesn't show that the theory is false, because of course it's impossible for it to be false. To claim that any theory is unfalsifiable is to actually claim that it is useless, which is not the same thing.

In the life cycle of a successful theory in orthodox science, there's an early stage where it gets tested. What that means is that the theory will be examined to see what kinds of testable predictions it might make, and then they'll be checked. Theories which make extremely surprising and even counter-intuitive predictions, whose predictions actually are found to be correct, are very highly thought of in science. Each such which is checked tends to add strongly to the conviction that the theory is right.

Eventually the confidence in it rises quite high and it begins to be applied. More or less, engineers adopt it and begin to use it as part of their process of making useful things happen. Instead of making predictions with it so they can be tested as part of verification of the theory, it's used to make predictions because those predictions are valuable and the theory is trusted to make them correctly.

It's a subtle point, but all useful predictions can be stated as a negative: "It is not possible that..." And the power of the theory is demonstrated by the breadth of what can be excluded. The more that the theory claims is not possible, the more useful the theory is – as long as it's right in the prediction. The most powerful theories make predictions of the form, "It is not possible that anything take place except for precisely thus-and-so."

Anyone can make positive predictions. There is a classic Johny Hart cartoon from the strip BC which has one of the characters reading a weather forecast which includes the possibility of rain, hail, blue sky, snow, hurricanes, tornados; the point being that such a prediction excludes nothing, and it cannot be converted into any kind of meaningful negative prediction telling us what cannot happen.

And unfalsifiable theories also can easily make positive predictions but cannot make negative ones. It cannot exclude anything. The claim of unfalsifiability is that the theory is stated in such a way that every conceivable outcome is consistent with the theory. But if so, then the only negative statement which can be made is "There is nothing which is impossible."

Theories are useful and trusted as a function of how good they are at making negative predictions, telling us what can't happen. But what's interesting about that is that this doesn't necessarily have anything to do with whether they're actually right. Newton's mechanics (which is often referred to as "Galilean mechanics" because Newton wasn't responsible for all of it) ruled classic physics for 200 years, but we now know that Newton's equations didn't give the right answers. They were really close in a lot of cases but not exactly right.

Einstein rewrote Galilean mechanics with the Special Theory of Relativity. However, within the realm of masses we humans can manipulate and speeds less than 1% of C, the difference between the values generated by the older equations and Einstein's rewritten ones are so small as to be unimportant, and as a practical matter we still use the older equations because they're simpler. However, in extreme cases their predictions can be substantially different, and experiments proved that Einstein's equations were better and must be used in the well known cases where the differences are great.

Are the Einsteinian equations "right"? We can surely not know. Perhaps someday we'll encounter even more extreme situations where Special Relativity (as augmented by General Relativity) may give us the wrong results, and then they'll have to be revised again. For the moment they're really, really good. (Though I've read that there's a way in which Relativity and Quantum Mechanics contradict each other, and experimentally QM turned out to be right. It's one of the interesting unsolved problems of physics to figure out how to resolve this.)

Even though Newton's mechanics was "wrong", it was damned useful and remains so today. Usefulness is an important characteristic of robust scientific theory, but it doesn't have anything to do with absolute truth or falsity as such.

The fact that the Theory of Fred is unfalsifiable and thus useless is unimportant for our purposes. If it were being proposed as a legitimate scientific theory, it would be rejected immediately because of its complete inability to create negative predictions. But we don't care about usefulness within this discussion.

The strong claim regarding atheism relates to proof, not to usefulness and the burden of proof is on the atheist to prove that it is false. If it actually is unfalsifiable, then it represents a counter-example to the strong claim regarding atheism because it cannot be disproved.

The proof atheist then asks the following hypothetical question: What if we can come up with a valid mechanistic explanation for the Big Bang? In that case it won't be necessary to refer to Fred as an explanation. We'll have two competing theories, and since the mechanistic one involves fewer assumptions then according to Occam's Razor it will be the right one, and the Theory of Fred will be wrong and thus be disproved.

Unfortunately, Occam's "Least Hypothesis theorem" is a heuristic. It states that when faced with several theories all of which explain the same data, you're best off going with the one which involves the least radical assumptions. It's really just a kind of guidance, a way of picking among them when there's no other way of doing so and we continue to use it primarily because it's been very valuable in the past. But like all heuristics it's possible for it to be wrong. (The old joke is that if it wasn't possible for a heuristic to be wrong then it would be an algorithm.)

Suppose that at the time of Newton both his mechanics and Relativity were simultaneously proposed. The scientists at the time would have been able to see how they made different predictions at extreme velocities but would have no way to check that, and both theories would make the same predictions for all tests they could perform, within the limits of experimental error. Occam's Razor would cause them to select Newton's equations over Einstein's (because of the assumption in relativity that C is identical in all inertial frames of reference), but that would have been the wrong answer.

Like any valuable heuristic, Occam's Razor is right most of the time but not invariably. As such, it can't be invoked as part of a proof. It's true that a mechanistic explanation involves fewer assumptions than the Theory of Fred, but that doesn't prove that the Theory of Fred is wrong.

Do I actually believe the Theory of Fred? Of course I don't; no one does. As a real religion it's ludicrous. Its only value is as a thought experiment, as a religious theory which contains no intrinsic contradiction and which cannot be disproved empirically. The strong statement regarding atheism that we are examining denies the validity not only of existing theistic religions but of all hypothetical ones as well even if they are preposterous ones which no one would actually ever believe. Because there's no way for us to differentiate between a totally mechanistic universe and the one created by Fred, it means that we cannot actually prove that the universe is mechanistic. All we are left with is opinion and belief and probabilities.

Which is to say, faith.

When I was younger I studied science and observed the real world and came to the conclusion that nothing I saw actually required a deity for explanation. And indeed I did apply Occam's Razor and concluded that the best guess was that no deities exist. I believe that there are none, but I make no claim to be able to prove it. On the contrary, I forthrightly state that I cannot.

It is possible in some cases to empirically disprove a belief. If I say that I believe that there is a God and that He lives in a palace on top of Mount Rainier, someone can take me up the mountain and show me that no such palace exists there. Beliefs can sometimes be disproved, but beliefs cannot be proved. (If they could be proved, they would be facts and not beliefs.)

And at about this point the theist speaks up, but that's going to have to wait for another post.

My new blog: Chizumatic