Stardate 20030518.1537

(On Screen): In response to a pair of posts I made last week, regarding the question of whether mechanism is a provable fact or a belief, The Raving Atheist has made a series of posts (1 2 3). In them, he makes certain statements about what he inferred from my writing about what I believe, in some cases quite incorrectly. For instance, he seems to have concluded that even if I don't agree with him that all religion is subject to outright disproof that I nonetheless do agree with him that Christianity specifically can be.

He seems to have been on the receiving end of a lot of criticism by third parties who have been referring to my posts, because he seems to go out of his way to try to claim that in almost all fundamentals he and I do agree. For instance:

It’s interesting how many bloggers who agree with Steven Den Beste on the topic of atheism disagree with me, even though there’s not the slightest practical difference between Den Beste’s atheism and mine.

As I pointed out in a previous post, Den Beste agrees that the existence of the Judeo-Christian god may be disproven. That god can be shown, depending upon the precise definition, to be internally inconsistent or empirically false, just like a square circle, an invisible pink unicorn, or the Wizard of Oz.

In the end, he dismisses the apparent differences between his point of view and mine as being due to my unwillingness to actually stand up for what I believe, thus appends the following:

To further fuel your righteous rage against Den Beste's traitorous, pandering conduct, read Jason Malloy's comments here and here.

I guess I can see how he might interpret what I wrote as pandering, but it's difficult to see how it could be traitorous. That presumes that I'm part of some formal group and had given some sort of promise to it of loyalty, and then turned around and broke that promise. But I am part of no such group. I do not owe any loyalty to any kind of global atheist cause.

After seeing all these claims about me, I felt the need to set the record straight and posted the following brief comment:

I do not agree with you that it is possible to disprove the existence of the Christian God. Nor do I "share your disdain for a tooth-fairy God" etc.

In response, Jason Malloy said:

Way to leave a comment completely lacking any sort of content SDB. If you're not going to challenge any sort of basic premise or logical platform to support your position what's the point of even saying anything?

No one cares what you agree with or don't agree with Den Beste, if you don't have any intelligent arguments to support yourself.

I was not in that comment attempting to justify my statement; I was trying to correct what I saw as mischaracterization of my point of view by RA. I simply wanted to include my demurral as part of the overall thread, for historical reasons. Malloy's comments are mocking, but in a sense his point is a valid one, and I think I'm going to have to make more clear the real source of the fundamental disagreement between RA and myself. What it comes down to is confusion about the difference between deduction and induction.

In many ways, deduction is the crown jewel of the body of human intellectual achievement. All of mathematics is based on it. It is formalized in the studies of sentential logic, set theory and Boolean algebra. Deduction permits actual proofs, and it is highly objective and repeatable. If several people start with the same body of axioms and procedures within a mathematical system and are asked whether a given statement is provably true within that system, then if they are skilled in the art they'll all arrive at the same answer. Of course, they may not all be equally skilled, and if it's a tough problem some may get answers and others may not, but if two such people arrive at contradictory answers than it means one of them made a mistake. If an answer in mathematics is right, it's all the way right; if it's wrong it's all the way wrong. Mathematics presents us with a virtual conceptual space in which is crystal clear and where "true" and "false" are absolutes. (Yes, I know about Gödel.)

Deduction is extremely powerful, and it can be applied to the real world in many important ways. Sentential logic can be used directly, and various other kinds of mathematics can be applied through isomorphism.

But for all its utility, deduction is subject to considerable limits, and there are many real world situations which are outside of its scope. For one thing, deduction cannot operate on insufficient data or on data of doubtful reliability.

In such cases, we fall back on induction. Deduction is a human invention (or discovery; opinions on that vary) but induction seems to be something we're actually bred to do. It's an inherent aspect of how our nervous systems work, and it can and has been demonstrated that other vertebrates far less sophisticated than we are use induction even though deduction and higher mathematics are beyond them.

If deduction is the realm of logic and proof, induction is the realm of the heuristic. Deduction is perfect logic; induction is good-enough logic.

Induction is not as reliable as deduction. It doesn't always get the right answer. But it can be applied to a far more broad set of circumstances than deduction, and it has a decent chance of providing a good answer in cases where deduction would be useless.

A heuristic is a guideline, a principle which is far more likely to be right than wrong. All of us maintain huge internal libraries of heuristics, the accumulation of a lifetime of experience, and for each we maintain an evaluation of how valuable it might be. People trade heuristics with one another; they share the ones they've found to be particularly valuable. Children absorb them from their parents. A lot of folk traditions and other kinds of orally-transmitted culture include large bodies of heuristics.

Among its many virtues, induction permits you to come to conclusions about how to operate in situations where you don't fully understand what's going on. The heuristic tells you what works, but not necessarily why it does, and you may be able to start to create useful heuristics long before you can explain them, if indeed you ever can.

Some heuristics are instinctive. Many people are aware of Pavlov's experiments in conditioning, but overestimate the importance of what he found. Pavlov was dealing with low level neural circuits, at a level little higher than reflex. Fewer people are aware of the work of B. F. Skinner, who studied forms of conditioning which appear to operate at the highest levels of cognition. What Skinner ultimately demonstrated is that vertebrates (including us) are wired to understand that correlation often implies the existence of causation. If you do something and get a good result, try doing it again.

That makes sense, of course. Places where food have been found are likely to have food if you check back. If you fly into the outdoor food court and land on the ground and cheep loudly, humans who are eating there may throw something edible onto the ground for you. Never mind why it works; the important thing is that it does – usually. It certainly works often enough so that the payoff if it does is more than worth the cost of giving it another try. If it keeps failing, eventually you discard it. But if it works often enough, you'll remember it and keep applying it. Extremely complex behaviors can be subjects of this kind of conditioning, well above the level of the salivation of Pavlov's dogs.

Skinner and others later studied this kind of conditioning closely, and have identified elements of the heuristic in pretty great depth. For one thing, it doesn't have to work every time in order to be reinforced; in fact it can be pretty rare. On the other hand, random payoffs are far more effective at reinforcement than periodic ones: the behavior is reinforced much more by a random 10% chance of payoff than by a payoff that happens every tenth time on a regular schedule. And if the payoffs vary in value, with more rare ones being bigger, that reinforces it even more.

It turns out that a slot machine (invented a century before Skinner's work) is a nearly perfect conditioning system to teach humans to feed it coins. (And that's why some people get hooked.)

Continuing, induction works by taking inadequate data of doubtful quality and assigning probabilities and reliability levels to what is known. It uses heuristics and experience from the past to try to fill in the gaps or to correct information which is likely to be false, and out of that it tries to yield conclusions which have the highest probability of being right, even if they're not certain. And then we act on those conclusions, and sometimes we lose because the answer was actually wrong. But if you refuse to act, you usually lose; inductive results are far better than nothing at all.

But all of this is highly subjective. The process of evaluating the reliability of each piece of data is based on prior experience and indeed may also be affected by the biology of that person's brain. Where deduction is objective, induction is not because it's based so strongly on personal experience and on prior conclusions and existing heuristics and biological differences.

All of us go through life constructing heuristics and storing them for later use; it's a constant process. We all constantly revise our valuation of those heuristics, and sometimes we discard them entirely. We actually need a pretty substantial body of them in place by the time we reach adulthood, but by their nature heuristics are also not necessarily perfect. (As the old joke goes, if a heuristic never failed it would be an algorithm.)

Some of the ones we use have been directly identified as "fallacies". And in fact they are, in the sense that they are not guaranteed to never fail. Some of the standard fallacies are failures of reasoning (such as begging the question) or deliberate attempts by the arguer to deceive (such as the strawman) but many of them are actually useful heuristics even if they're not totally true. Post Hoc Fallacy is Skinner's conditioning. Appeal to authority can lead you astray, especially if the authority is commenting outside his field of expertise, but as a practical matter it's worth listening to what wise men say, and we generally tentatively assign a high value to what such people tell us. Part of why we make these mistakes is that a lot of the time they actually give us good results.

Superstitions and prejudices are heuristics, though mistaken ones. The process of using Skinnerian "it works even if we don't know why" to deal with complex situations is actually fruitful quite a lot of the time. But it, too, can lead us astray and that's how a lot of superstitions develop. I remember seeing an old man playing the slots in Vegas who would rub his hand over the front of the slot machine while the rollers turned. Every time, in fact. Why? Once he did that and hit big, now he does it every time. It doesn't affect the slot machine, but he thought it might since it seemed to once.

Superstitions arise because of a mistaken belief of causation in a case which was actually coincidence. They're the pathological demonstration of the imperfection of Skinnerian logic; it's useful a lot but isn't invariably right. I understand the internal construction of the slot machine, and the laws of probability (and, not incidentally, also the laws of the State of Nevada regarding gambling) and thus I know full well that what he did had no effect. (It better not, or the Nevada Gaming Commission is going to be all over that casino.) But he didn't know those things, and for him it was a matter of trying to cope with a situation without full knowledge.

Prejudices often arise the same way. In the course of our lives we meet people and get to know them, and some of them turn out to be nice and friendly and some of them turn out to be dangerous or unpleasant. Some encounters are brief, some long lasting; some are benign, some are immensely valuable, and some are horribly damaging and may even involve physical danger. And we develop heuristics about this just like we do everything else, and try to judge each new person we meet on the basis of past experience so as to try to make a preliminary evaluation of how likely it is that this person will be worthwhile to know or someone that might be dangerous.

Like it or not, that's based on rather surface evaluations. And like all heuristics the result is error prone. But though philosophically I understand and agree that we should all try to treat every person we meet as individuals, if you think about it it's clear that some kinds of characteristics which we may fear do tend to correlate with other features we can directly observe.

Which is to say that it is not the case that every person who fits a certain "racial profile" is certain to be a crook, but they are more likely to be, and that's the kind of level that heuristics operate at. Some such heuristics are totally wrong, but some are actually based on valid statistical evaluation.

Right or wrong, we all do this to some extent. What we refer to as prejudice is cases where heuristics like this get out of control, and where people rely far more heavily on them than they should. And in fact, any heuristic becomes damaging if someone gives it more credence than they should; the reason we study those logical fallacies is that people treat them as forms of deduction (and certainty) instead of forms of induction (and persuasion).

Induction can operate at many levels of sophistication, and it can be implemented at a level of practical usefulness with far less compute power than exists in the human brain. It's the level on which animals think; they don't worry about why things are happening; they just observe what works and what doesn't. It's a cheap way to get pretty good answers most of the time.

Of course, if one applies more data to the inductive process, one can sharpen the result more. Humans vary in intelligence and in knowledge, and those on the right side of the scale will often arrive at quite different evaluations of the reliability value of different conclusions. Over on the left side of the scale, inductive processes use less data and less computation and tend to rely on fewer and more broad heuristics which are more prone to be incorrect. Though it yields results, the results are more likely to be erroneous. Induction increases in reliability as a function of how much data and experience and computational power are applied to it, but induction is never certain. If it were possible to actually be completely certain about a subject, we'd be using deduction and wouldn't need induction at all. (Of course, some people not familiar with deduction or without adequate access to data will use induction in such cases anyway, and as a result won't necessarily arrive at the correct answer.)

All of us use induction constantly, but most of us don't really realize it and aren't aware of what is involved. Induction doesn't offer us conclusions, in the sense that deduction does. What it gives us is a feeling of how likely it is that a given statement can be relied on, ranging on a scale from "total horseshit" on one end to "virtually indistinguishable from fact" on the other end. But you can often tell when someone is offering you an inductive conclusion rather than a deductive conclusion. If they say something like "It just makes more sense to me that..." what they're saying is that their inductive calculation arrived at a high reliability for that statement. If someone cites something as an intuition, they're talking about inductive conclusions.

If they make arguments by citing pieces of evidence that they found persuasive rather than by making actual deductive proofs, then they're talking about induction; persuasiveness is the essence of induction because it's all based on probabilities and likelihoods. What they're doing is to show some of the input to the process which they ended up giving strong weight. Of course, if someone else doesn't evaluate the weight the same way, they in turn may not find that argument persuasive at all.

Those on the left side of the scale often do arrive at poor results using this process, but it is not the case that everyone on the right hand side will arrive at the same answers.

Which brings me to the key point: I am totally convinced that a mechanistic explanation for the universe is the correct one. But I arrived at that result through inductive reasoning, not through deductive reasoning. For me, mechanism is at the "virtually indistinguishable from fact" point on the certainty scale. I arrived at that point in my mid-20's, and 25 years later it's only become stronger and stronger.

But induction is not objective. Deduction is objective, and one can document the evidence, reasoning and conclusion of a deductive proof and others can look at it and will agree that it's unflawed and therefore true. Induction is not ultimately susceptible to an equivalent process of checking.

Which is why humans have, in the last three hundred years or so, developed a metasystem on top of personal induction which is designed to conserve the best conclusions induction offers us while having the best chance of rejecting and discarding its mistakes. It's a sort of collective and collaborative inductive process called "the scientific method". For instance, if there are alternative hypotheses about something, proponents of each may try to argue for why the data others relied on might be less reliable or persuasive than previously thought. Or perhaps someone may be able to propose a way of collecting and evaluating new evidence which would strongly support one theory over another. It's been highly successful, and if mathematics is the triumph of deduction, science is the triumph of induction. It consists of a body of theories which are not provably true (in the mathematical sense of "proof") but which in the best cases are so inductively certain as to be indistinguishable from fact. On the other hand, there are other theories within that overall body whose reliability is open to serious doubt, such as the current state of predictive climatology, or which are inductively at the "total horseshit" end of the scale, such as "creation science" or the ether theory of light or the phlogiston theory of combustion.

In some cases these are theories which at one time were respectable, which is to say that on the basis of inductive evaluation of what was known at the time, they seemed to be plausible. All three of creationism and the ether and phlogiston were once central theories in science, but in each case new data was developed which dramatically reduced the consensus evaluation of their reliability, and no one now in mainstream science takes any of them seriously.

But in some cases we as individuals can arrive at equal degrees of certainty about some issues but without really having any ability to convince other people of those results, in fields which are not readily susceptible to the collective processes of the scientific method. Religion is one of them.

I have enormous respect for Donald Sensing. He's about my age, and he is extremely sensible. He is analytical and intelligent and very well read. He has a great deal of practical life experience. He was a career officer in the military, serving in the artillery and doing staff and intelligence work. Donald is no fool.

Which is why it's noteworthy that on the subject of religion, the two of us arrived at such diametrically opposite conclusions. I am absolutely convinced that mechanistic atheism is correct, but Donald believes so deeply in his Christian faith that after he left the Army he went to divinity school and was ordained as a minister, which shows a degree of commitment and certainty beyond that of the normal believer.

In private email correspondence, Donald and I jest with each other about this difference, but neither of us has ever made any kind of serious attempt to influence the other, in part because I think we each respect the other enough to know that it would be fruitless.

Each of us arrived at our conclusion inductively, and though the body of relevant knowledge we each have almost certainly isn't identical, there's considerable overlap, and there's no particular reason to believe either of us has absorbed a significant body of false information, or that either of us relies heavily on any body of information that the other has no knowledge of. Donald isn't one to make stupid mistakes about misapplication of the Second Law of Thermodynamics; he couldn't have been an Artillery officer without a decent understanding of Physics.

And I can see the kind of reasoning process he uses in some of his posts where he analyzes ethical issues. He starts with different axioms than I use, but beyond that point there's little to criticize. If I agreed with his axioms, I'd also agree with his conclusions. The intermediate reasoning is totally sound.

In deduction, if you start with different axioms, you can arrive at different results but if you start with the same axioms then if no one makes a mistake you'll end up with the same result. But that's not true in induction.

If two people have different bodies of knowledge, they obviously won't necessarily get the same inductive answers. But even if they broadly share the same knowledge base they can still get different inductive results.

That's because they may place different weights on different pieces of data. What one may find critically persuasive the other may think is completely unimportant. In principle that whole process might be subject to extreme scrutiny and maybe everyone could come to agree on all of those evaluations, but as a practical matter it ain't gonna happen.

That's why Donald and I don't agree about religion. We both know more or less the same things, but we evaluate different parts of that knowledge as being critical to the inductive conclusion. (This is, of course, my mechanistic explanation for it all. Donald would say that he has had a spiritual enlightenment and I have not, that he is open to hearing God speak and I have closed my mind to that and refuse to listen. But that amounts to using different words and concepts to say the same thing: what influenced Donald was different than what influenced me, even though both of us have been exposed to all of it.)

Going back to my post on the question of "proving atheism", what I hope I can now make clear is that I am absolutely convinced, to the level where I harbor no lingering doubts at all, that a mechanistic explanation for the universe is the right answer. But it's the result of induction, not deduction, and inductive conclusions are not "proved". The fact that I'm certain doesn't mean I can prove it to anyone else.

Oh, I can explain some of the kinds of things I've seen which I think are highly persuasive, and which strongly influenced my evaluation of the issue, and in fact I've done so on this site. But if someone else doesn't assign high weight to such things, they won't be convinced by that. And unless there's a practical way to test inductive conclusions, then ultimately there's no way to arrive at consensus.

When The Raving Atheist tries to claim that atheism is true and that theistic religions are false and that these things can be proved, those claims only make sense within deductive reasoning. Induction can never prove anything; all it can do is to assign an extremely high conviction level to it. What I was trying to say in that article was that my certainty about atheism is based on induction, and because induction is subjective I cannot outright prove to anyone else that I'm right.

If someone makes the claim that atheism is subject to deductive proof then they are forced to achieve the standard of proof that I described in that article, and any single counter example, even if preposterous, is enough to derail that claim. That is the nature of the deductive process; it's what makes it powerful when it works. If no such counterexample can be found, even preposterous ones, then the original claim is very powerfully proved.

But what I showed in that article was that it is not in fact subject to deductive proof, which is why my single preposterous example was relevant.

On the other hand, what works to disprove a claim of deductive truth may be irrelevant when the conclusion is the result of induction. Which is why this is highly revealing:

A number of bloggers who have commented on his posts seem to have missed this point, not realizing that Den Beste’s assertion that atheism and theism are both a matter of “faith” only applies to trivial, meaningless constructs such as Fred or Balder in which nobody believes and which Den Beste himself calls “ludicrous. Devout Christian Tertius calls Den Beste essay “a very sensible, well-reasoned and non-inflammatory piece.” Frank of IMAO, also a believer, calls Den Beste’s piece a “nice essay.” Rachel Lucas, an atheist who is apparently nonetheless sympathetic to her father’s Christianity, calls Den Beste is a “genius.” The confusion seems to stem from Den Beste’s assertion, quite mistaken, that the failure of atheism as to Fred implies the failure of atheism as to the Judeo-Christian god.

What he's saying is that my "Theory of Fred" is specious and uninteresting, and within the realm of inductive reasoning he's right. And what becomes clear in reading his comments on this whole issue is that he, like me, arrived at atheism as an inductive result of very high reliability, but unlike me he doesn't understand that induction is not proof. And when he presents his inductively persuasive arguments to others and they don't respond as he expects, he contemptuously dismisses them as deluded fools.

He makes a claim several times that he and I agree that the Christian deity specifically can be disproved. No, I do not agree with that.

One reason why is that as a practical matter, there is no single Christian deity. "Christianity" refers to a collection of religions, but within that collective there is enormous disagreement and in the past some Christians have declared others as heretics because of those disagreements and have even gone to war about them. There is no single "Christian God" except in an uninteresting rhetorical sense that just about the only thing they all agree on is that there is one. But as to His characteristics and properties and attitudes and intentions and powers and behavior, there's no consensus whatever.

For that matter, there's even debate about which sects are actually Christian. I've met Christians who don't think Mormons are Christian, for instance. On the other hand, Unitarianism used to be considered Christian though few think of it that way now.

It is possible to evaluate specific strains of Christianity, and in some cases to deductively disprove them. For instance, there are sects which claim to believe in Biblical inerrancy: that the Bible is the literal Word of God and that it is absolutely true in every regard.

That claim is actually subject to direct disproof. There is, for instance, a description in the Old Testament of a very large round object which was created for some reason-or-other, and both its diameter and circumference are given (in units of "cubits"). The ratio of the stated circumference and the stated diameter is exactly 3.0, and in fact if the stated diameter was right, the stated circumference was off by a couple of cubits; no rounding error explains the numbers. That object could not have had the measurements it is claimed to have had unless it was hexagonal instead of circular.

Even if one discounts that or rationalizes it away, there are places in the Bible where the same events are described in more than one book, and the descriptions contradict one another. Which means that they might both be wrong but they cannot both be right. (I used to have a really long list of such things but I don't know what became of it. Doubtless something like that can be found online somewhere, but it doesn't really matter.)

But other flavors of Christianity don't require Biblical inerrancy, and adopt the idea that the Bible was written by humans but inspired by God. In that case, the mistakes come from men and don't impeach the book overall, and claims of internal contradiction or factual error are uninteresting; they already have an explanation for them which is consistent with their faith.

Some Christian faiths can actually be deductively disproved in ways like this, but such disproofs will have little effect on other Christian faiths. The collectively "Christian God" doesn't exist because the differences among the faiths easily are more important than the similarities and there's not enough consensus to provide material for any kind of single deductively conclusive argument.

On the other hand, one can try to present an inductive case against the Christian God which is overwhelming. The Raving Atheist thinks such a case exists, and so do I, which is part of why both of us are atheists. But what I understand and RA clearly does not is that such an argument isn't objective.

As a practical matter, it's a belief. I know that, and say so. He doesn't seem to realize it.

In induction what one person finds overwhelmingly convincing, another may find weakly interesting if at all, and this doesn't necessarily mean one is right and the other is wrong. Sometimes they are, but sometimes it just means they're evaluating the importance of the issues involved in different ways.

I know or have known Christians that I thought were deluded fools. I have received more than my share of incoherent arguments from them which they thought were overwhelming and I thought were clear demonstrations of their lack of education and inability to reason effectively.

But I have also known, and now know, Christians who are intelligent, intellectually disciplined, well read and knowledgeable who are completely convinced of their faith, just as I am totally convinced about mine. I think they are wrong, and they think I am wrong, but it isn't possible for either of us to prove it to the other. These are not people who believe in a 7-day creation 6000 years ago; these people believe in an ancient earth and think that humans are the end result of evolution, but think that there's more to the universe than I do. The claims that they make are subject to inductive evaluation but not to deductive examination or empirical test, which means that we can personally arrive at conclusions about them but cannot objectively prove those conclusions to each other when we end up disagreeing about it.

I think that a Christian who believes in a 7-day creation etc. is an idiot, but I do not think Donald Sensing is an idiot, and I consider his religion to be equal to my own, even though it totally contradicts mine. That doesn't mean I have any doubt about mine; I'm convinced he's wrong.

Not all religious positions are equally worthy of respect and many are only worthy of utter contempt. But I do not grant that the atheistic point of view is in any way above all of the best of the alternatives, such as the one Donald believes. Donald and I disagree, but I think that his inductive process was of equal quality to my own, and until such time as further evidence becomes available which might directly affect those calculations, we'll continue to disagree while maintaining mutual respect.

As a practical matter, if he's right we'll both find out when we die. If I'm right, neither of us will ever know for sure. But absent some new round of miracles, such evidence won't appear while we're alive here on Earth.

My new blog: Chizumatic