My last post dealt with some technical problems with specifying the probabilities for the fine tuning argument, using part of chapter 7 from Alvin Plantinga’s Where the Conflict Really Lies as a starting point. Partly due to the technical nature of the material and partly due to lapses of clarity in my writing, the post will likely be opaque to readers new to this topic. A reader suggested that I write an introduction to the subject in layman’s terms. Since I’m probably not the best qualified to do this — I’m a philosophy teacher, not a physicist, Jim — allow me to recommend a few resources for those interested. The book that got it all started is John Barrow and Frank Tipler’s The Anthropic Cosmological Principle. I haven’t read it myself, but it’s cited in just about every other book or article on the subject that I’ve read. Other good books on the subject are: Just Six Numbers by Martin Rees, The Goldilocks Enigma by Paul Davies, and Universes by John Leslie. If you’d prefer to listen to a podcast that summarizes the material well, check out this interview with astronomer Luke Barnes.
The Fine Tuning Data
In what follows, I’ll try to present the fine tuning argument (FTA) and several of the popular responses to it. These responses are not all created equal, so I will deal with them under the headings ‘bad’, ‘better’ and ‘best’. But before we can get the argument on the table, we need to ask ‘what is fine tuning?’ Fine tuning is the term given to the observation that a number of physical constants — basically, the values of features like the strength of gravity and the charge of the electron — need to fall within a narrowly defined range if the universe is to permit life at any point in its history. For example, if the universe had expanded at a different rate, life would not have evolved. According to John Leslie, had the universe’s expansion decreased by one part in a million million, the universe would have collapsed. Had the universe’s expansion increased by one part in a million million, there would be no stars, galaxies, or planets and, thus, no life. Or had the gravitational force been slightly greater, all the stars would be blue giants unable to support life. But had the gravitational force been slightly less, the universe would lack many of the elements necessary for life. Had the charge of the electron been slightly different, stars would either have been unable to burn or wouldn’t have exploded and produced heavy elements necessary for life. Examples like these could be multiplied, but I think that the point has been made. The inference from this data, then, is that it’s highly improbable that these features of the universe would have just the values they do by sheer chance.
The Fine Tuning Argument(s)
We should be careful to distinguish the fine tuning argument from the fine tuning data. The FTA is not the same as the fine tuning data. The data, from what I can tell, is relatively uncontroversial. The fine tuning argument takes the data as a starting point and draws a theological conclusion from it. One of the difficulties with talking about the FTA in the singular is that there are different ways to set-up the argument. I’m going to differentiate two versions of the argument: deductive and inductive. The deductive version is probably the most straightforward. I’ve seen William Lane Craig defend a version like this: P1. The fine tuning of the universe is due to either chance, necessity or design. P2. It’s not due to chance or necessity. C. Therefore, it’s due to design. Of course, a lot more work would have to be done to defend the premises, but if they are true the conclusion follows. The other way one can set up the argument is as an inference to the best explanation. On this interpretation, the fine tuning of the universe confirms theism as a hypothesis. In other words, if we have two hypotheses, theism and atheism, fine tuning is better explained given theism than given atheism. This can be worked out in greater detail, and Robin Collins is probably the best proponent of this version of the argument. The way one responds to the FTA depends in large measure on the type of argument presented, but I want to survey a few general responses. I’ll try to keep them brief. A lot more could be said about each one, but I leave it to interested readers to pursue these discussions by leaving comments.
Bad Responses to the FTA
1. One response to the fine tuning data is to say that they are just coincidences. The idea here is that any values of the physical constants would be equally improbable, so the apparent fine tuning requires no special explanation. Somebody has to win the lottery after all. The problem with this response is that there seem to be too many coincidences. Moreover, as far as we can tell, these values are independent of one another. In order to permit life, the universe would have to win multiple, independent lotteries and many will think the odds against it too high.
2. Another response is to say that although the universe appears to be fine tuned for human life, there may be other forms of life that could have evolved even if the universe had been significantly different than it is. However, this response largely misses the point of fine tuning. Regardless of what form of life one envisions, life needs complex chemistry. You can’t build intelligent life, of any kind, on hydrogen which is the only element the universe would contain if certain parameters were different. Thus, the ‘other forms of life’ response is fundamentally misguided.
3. The final response in this category is to say that the universe, on its face, doesn’t seem that finely tuned for life. There is, after all, a lot of empty space. If God intended the universe to be teeming with life, he clearly did a poor job. Again, this objection largely misses the point. Fine tuning does not mean that the universe contains as much life as it possibly could. Rather, the claim is that certain conditions need to be met in order for the universe to contain any life at all. In fact, the size and age of the universe are part of the fine tuning data, so this response is simply misinformed.
Better Responses to the FTA
1. A better response to the FTA is the so-called Anthropic Principle. This principle is articulated in various ways, but basically it points out that a necessary condition for observing that the universe is life permitting is that the universe is, in fact, life permitting — otherwise, we wouldn’t be here to observe it! This is self-evidently true, but a little more work needs to be done before it can properly be called an objection to the FTA. With the Anthropic Principle in mind, some critics of the FTA (Elliot Sober is probably the best), suspect that there is an observational selection effect (OSE) at play in the argument. Again, we couldn’t possibly observe a universe that does not permit life, so we couldn’t fail to find the evidence that we do. The following analogy, proposed by Arthur Eddington, is sometimes used to elucidate the alleged fallacy of the FTA. Suppose you are fishing with a net which has a very wide mesh such that it only allows you to catch fish over ten inches long. Obviously, you’re going to observe fish over ten inches long, but it would be wrong to conclude on this basis that every fish in the sea is over ten inches long. That’s because your net produces an OSE. Likewise, so it goes, it would be wrong to conclude that the universe is ‘just right’ for us observers who, we may say, just happen to exist because the universe is the way it is. But, as Alvin Plantinga points out, not all OSE’s are fallacious. For example, the claim ‘I am sometimes awake at 3:00 AM’ is true even though I couldn’t observe this if I were not awake. Moreover, the FTA differs from the fishing analogy in an important way. The FTA does not reach a conclusion about what proportion of universes are fine tuned. Rather, the FTA attempts to reach a conclusion about the probability of this particular universe being fine tuned. A more apt analogy is the oft-cited firing squad analogy from John Leslie. Let’s suppose you are facing a firing squad consisting of expert marksmen. They fire at you but, much to your surprise, you have survived unharmed. Now, obviously you wouldn’t be able to observe that you’re alive if you had been killed, but this is hardly relevant. You can still form a judgment about the probability that the expert marksmen accidentally missed or that they missed intentionally. So while I think that we should affirm the self-evident Anthropic Principle and be wary of OSE’s in certain contexts, I’m not convinced that these considerations are enough to constitute a strong objection to the FTA.
2. Another decent response is to say that, for all we know, there is a deeper law of physics that would make the values of these constants physically necessary. In other words, perhaps these constants are not simply independent coincidences — or the product of design — but are bound together by a more fundamental law of nature, perhaps the much sought after ‘theory of everything’ or ‘grand unified field theory.’ If we discover such a law, it may turn out that the constants of nature simply must have the values that they do. Indeed, nobody knows what the future of physics holds, but this speculation by itself is more of a promissory note than an objection to the FTA. Some theistic critics I’ve read have called it ‘naturalism of the gaps.’ Although I think they overstate the point, I take it for what it’s worth. Another potential problem is that a deeper law, even if found, may exhibit fine tuning of its own. In that case, we haven’t really explained away fine tuning, but merely pushed it back.
3. Finally, there is the popular multiple universes, or ‘multiverse’, objection to the FTA. The multiverse objection concedes that the fine tuning data is improbable provided the number of actual universes is 1. However, there may be many, perhaps an infinite number, of actual universes each with different values for the fundamental constants of physics. With a sufficiently large sample size, it no longer seems so improbable that some of these universes would, by chance, have life permitting values. And, of course, we are in one of the lucky universes that permits life. I suspect that this response, in combination with the Anthropic Principle, is the most widely invoked response to the FTA (See Richard Dawkins’ The God Delusion). There are advantages and disadvantages to this response. It’s advantageous because it seems to be in vogue in physics at the moment to speak of multiple universes to explain a range of strange phenomena. Thus, it’s more difficult to accuse it of being an ad hoc response to the FTA or an example of the gambler’s fallacy. It’s disadvantageous because there is no direct observational evidence for it, nor does it seem that there could be. As such, theists typically accuse the multiverse of being as metaphysical, and less simple, an explanation than God. I would contest both charges, although to do so in detail would take us far afield. I’m just anticipating the likely course the dialectic will take once one invokes the multiverse response.
The Best Responses to the FTA
The best responses to the FTA, in my judgment, are those that question the way that FTA proponents assess the relevant probabilities. Unlike some of the responses above, this tack does not attempt to offer an alternative explanation of the fine tuning data. Rather, the object is to show that the FTA doesn’t succeed in establishing its conclusion. Although the atheist might like to have an explanation, all he really needs to do is show that the argument doesn’t work (or, in the case of the inductive version, doesn’t work as well as the FTA proponent thinks). Because of the focus on probability, this approach tends to be rather technical, but I’ll do my best to stick to layman’s terms.
1. The Infinite Probability Space Response (IPSR) is the term I will use for what’s called the normalizability objection. I think IPSR captures the idea better for most people. This objection says that the FTA cannot be formally stated because the probability that, say, the gravitational force would have just the value it does, doesn’t add up to 1. In probability theory, probabilities, in order to be meaningfully stated, must add up to 1. For example, if you roll a six sided die, the probability that you’ll roll a 2 is 1 in 6. But, the probability that you’ll roll a number between 1 and 6 is 1 in 1. In other words, it’s certain that you’ll roll a number between 1 and 6 because all of the probabilities add up to 1 (1/6 x 6 = 1). So in probability terms 1 just means certainty. With respect to the FTA, the IPSR claims that the probability of gravity having the specific value it does cannot add up to 1 because the probability space is infinite. In other words, there is an infinite range of values that gravity could have taken. As such, we can’t say what the probability of its actual value is, which is just to say that we can’t meaningfully state the probability of the fine tuning data.
2. There is another way that one could argue that the probabilities are inscrutable. One could set up the FTA in Bayesian terms (Bayes’ Theorem is the standard way to calculate probabilities). The first step in constructing a Bayesian version of the FTA is to determine the antecedent probabilities of both theism and fine tuning. This simply means that we ask ourselves “what’s the probability of theism prior to the evidence of fine tuning?” and “what’s the probability of fine tuning prior to the evidence that the universe is in fact fine tuned?” The problem, of course, is how to estimate these probabilities. The theist and atheist are going to disagree about the prior probability of theism. And what’s the prior probability of fine tuning (again, independent of the evidence for fine tuning)? How could we know? I think the best we can say is that the probabilities are inscrutable.
3. Some proponents of the FTA realize this problem and propose a likelihood version of the argument. In this version, we disregard the antecedent probabilities and simply ask, “is fine tuning more likely given theism or given atheism?” In this case, we take the fine tuning data for granted and ask which hypothesis explains it best. This approach sounds intuitive, but it has problems of its own. First, it takes the fine tuning data as a given and then projects it onto our expectations about what God would or wouldn’t do. Assume that theism entails that God wants there to be life. How do we know that God wouldn’t create a universe with very wide life permitting parameters? Why wouldn’t we expect “coarse tuning” given theism? Second, if we take our background knowledge of fine tuning for granted, it implies that God couldn’t have created life any other way. In this case, the probability of fine tuning given theism is 1. It’s not just likely, it’s certain. But surely the theist doesn’t want to make so strong a claim. Moreover, if we can help ourselves to our background knowledge about fine tuning, what’s to stop us from helping ourselves to other background knowledge, namely the fact that we exist? Once we include this data, the probability that our universe would be fine tuned becomes 1 on either theism or atheism. This presumably isn’t the conclusion the theist wants. To clarify, I don’t think the probability is 1 on either hypothesis. I’m saying the probability is nearly impossible to calculate. The above is just an illustration that you can play around with probabilities to reach counter-intuitive conclusions.
Another problem with the likelihood version is that it produces an “arms race” with respect to competing hypotheses. Alvin Plantinga makes this point in his most recent book. Essentially, the theist needs to say that theism (at a minimum) entails that there is such a person as God and he wants there to be life. The atheist can say that there is no such person as God, but the universe has an intrinsic impulse toward life. The theist can add that God ‘really, really’ wants life. The atheist can respond in turn, and so on. If this looks silly, that’s because it is. Once you throw out the Bayesian interpretation and disregard antecedent probabilities, there is nothing to constrain this arms race of hypotheses. Normally, on a Bayesian interpretation, the antecedent probability that you assign to a given hypothesis would constrain this kind of unchecked modification of the hypothesis. But on the likelihood version, you can’t do that. As a result, you get a very weak argument, both probabilistically and rhetorically.
I think the FTA is a fascinating argument and one of the better weapons in the apologist’s arsenal. Unfortunately, its strength is often overstated. Hopefully, I’ve given a fairly lucid account of my reasons for questioning the argument’s power. There are a range of responses (some bad, some better, some best) that the non-theist can give as counter-arguments. The most promising tack, in my judgment, is to ask the theist to state the argument rigorously in probabilistic terms. I think once such a version is on the table, we can make some progress critiquing the argument.