This is the weekly Q & A blog post by our Research Professor in Philosophy, Dr. William Lane Craig. This week's response is the second part of Dr. William Lane Craig's answer to a question that appeared in last week's Q & A blog post; you can view the first part of the response here.
Thank you so much again for the spirited and great debates that you participate together with cosmologists. I have to say that I find this arena of Cosmology and God to be a very good ground to keep polishing primarily because when science proves a theory or finds very good evidence of it we come closer and closer to understanding more of that vast amount of matter and energy we can observe.
In any case I wanted to ask you about the latest debate with Sean Carroll. There were some strong points made in that debate that as a layman in cosmology make me want to seek further and further what are the theoretical physicists really saying on their theories. The media is not always clear on separating the cosmologist opinion/belief vs what their theory actually says without bias. So I went ahead and looked at Sean Carroll's post debate comments.
It seems to me that the objections that make me question the way the Kalam argument works as well as what the latest theories are showing vs cosmologists own opinion are Sean's answers to:
- First Premise of Kalam Argument (Aristotelean Causation). He digs deep onto Aristotelean analysis of causation being outdated.
- Boltzmann Brain problem. He mentions that the BB problem helps isolate those models of multiverses that are not tenable. So what about the models that do work?
- Fine Tuning. Sean mentions 5 points about Fine Tuning not being good argument for theism and he goes on and says that you did not respond to them. I understand that on your Q&A 49 you mentioned that the universe is fine-tuned for the existence of intelligent life is a solid fact; so I get confused with what Sean says about "he didn't offer any suggestion that we actually do know the conditions under which life can and cannot form."
Anyways, I hope you can answer this questions for me and help me advance in my pursue to understand modern cosmology, its facts and its hopes. Also, will you also make contact with Allan Guth about the BVG theorem? I find it puzzling that the first one to say you were wrong about BVG was Dr. Krauss and then we saw your e-mail exchange with Valenking proving otherwise. It would be nice to see for us viewers what your interaction would be with Allan Guth, now that Sean Carroll is claiming you are mistaken in the interpretation of the theorem.
Dr. William Lane Craig’s Response
As I explained in my last post, I waited until the transcription of our debate was complete so as to be able to comment responsibly on our excellent exchange. With the transcript now available for study, I dealt with Jahir’s first question in QoW #368. Today I want to turn to his second question, the Boltzmann Brain problem.
Situating the Problem
In order to situate the Boltzmann Brain issue properly in the context of the debate, I provide here an outline of my affirmative case in support of the proposition under debate:
I. Kalam Cosmological Argument: “The universe began to exist.”
1.0 Evidence from Expansion of the Universe
1.1 Classical spacetime began to exist
1.11 Borde Guth Vilenkin theorem
1.12 Carroll’s model ruled out on other grounds
1.2 Quantum Gravity era began to exist
1.21 If such a state did not have a beginning, it would produce classical spacetime either from eternity or not at all. So if such an era exists, it is the beginning of the universe.
2.0 Evidence from Thermodynamics
2.1 Universe began to exist in a low entropy condition
2.11 Carroll’s Model
2.111 Carroll’s model violates unitarity of quantum physics
2.112 Carroll’s model doesn’t solve Boltzmann Brain problem
2.2 Quantum Gravity era began to exist
2.21 Wall theorem holds for Quantum Gravity era and requires a beginning. One can avoid it only by having reversal of time’s arrow, which implies a thermodynamic beginning of time and the universe.
II. Teleological Argument: “The fine-tuning is not due to physical necessity or chance.”
1.0 Fine tuning is not due to Physical Necessity
1.1 M-theory permits 10500 different universes with various values of the fundamental constants and quantities
2.0 Fine tuning is not due to Chance
2.1 Multiverse hypothesis (incl. Carroll’s model) faces Boltzmann Brain problem
As my red highlighting reveals, the Boltzmann Brain objection actually occurs at two points in my case: first, in the context of the kalam cosmological argument as an objection to Carroll’s attempt to escape the implication of the second law of thermodynamics that the universe began to exist (2.112); and second, in the context of the teleological argument as an objection to the multiverse defense of the alternative of chance as an explanation of the universe’s fine-tuning (2.1). This double appearance is highly unusual, since one would normally give independent arguments in support of one’s points—not to mention the fact that one’s listeners can easily become confused as to which point is under discussion! That I would make the Boltzmann Brain objection do double duty in my case is a measure of just how truly formidable a problem it is for views unlucky enough to confront it.
Boltzmann Brains and the Thermodynamic Evidence for the Beginning of the Universe
In my opening speech, after explaining how Carroll attempts to avert the beginning of the universe by means of his time-reversal model of the universe, I object:
Professor Carroll’s solution provides no convincing answer to the Boltzmann Brain problem. Since the mother universe is a de Sitter space in which thermal fluctuations occur and since baby universes grow into de Sitter spaces themselves, there’s no explanation in the model why there exists a genuine low entropy universe around us rather than the mere appearance of such a world, an illusion of isolated brains which have fluctuated into existence out of the quantum vacuum. These and other problems make Professor Carroll’s model less plausible than the standard solution that the universe began to exist with an initial low entropy condition.
People who have not read Carroll’s book From Eternity to Here will not realize the sense of desperation that attends Carroll’s effort to solve the problem of the initial low entropy of the universe, which points to its beginning. Throughout the book he is driven back and back, until finally, when “We almost seem to have run out of options,” caught “in a viselike grip of logic, with no remaining avenues to reconcile the evolution of entropy in our observable universe with the reversibility of the fundamental laws of physics” (p. 349), his back apparently to the wall, he comes up with his remarkable model as his only solution to the Boltzmann Brain problem (pp. 361-3). This model represents his best shot at avoiding the problem; if it doesn’t work, other solutions will be implicitly even more implausible.
This version of the multiverse will feature both isolated Boltzmann brains lurking in the empty de Sitter regions, and ordinary observers found in the aftermath of the low entropy beginnings of the baby universes. Indeed, there should be an infinite number of both types. So which infinity wins? . . . Ultimately, it’s not enough to draw fun pictures of universes branching off in both directions of time; we need to understand things at a quantitative level well enough to make reliable predictions. The state of the art, I have to admit, isn’t up to that task just yet. But it’s certainly plausible that a lot more observers arise as the baby universes grow and cool toward equilibrium than come about through random fluctuations in empty space (pp. 363-4).
Carroll makes no attempt to justify his optimism. So my challenge to Carroll is to give us some explanation of why, on his model, our observation of a low-entropy universe around us is veridical rather than illusory. I claim that his model doesn’t give any answer. The mother universe pictured in Figure 7 of the transcript is a so-called de Sitter space in which fluctuations are said to spawn baby universes, and each of the baby universes expands until it becomes a de Sitter space like the mother universe, spawning more babies. So why think that more of these fluctuations produce genuine low entropy universes rather than much more probable observers who are mere Boltzmann Brains having an illusion of a low entropy universe around them? Carroll might say, “But more and more baby universes with ordinary observers will continue to form!” Ah, but those observers are in other universes (remember: the babies pinch off to become separate spacetimes)! In this, and any, baby universe, the Boltzmann Brains will in the long run vastly dominate as ordinary observers become extinct. And even if it were true that inflationary patches spawning baby universes are more probable than fluctuations forming Boltzmann Brains (itself a moot point), that still fails to explain why we find ourselves surrounded by a low entropy condition.
Notice that I have now put the ball in Carroll’s court. Watch to see if he provides the requested explanation.
Boltzmann Brains and the Chance Explanation of the Fine-Tuning of the Universe
Later in my opening speech I return to the Boltzmann Brain problem when discussing the attempt to explain the fine-tuning of the universe as the result of chance. Proponents of the multiverse hypothesis (like Carroll) try, in effect, to multiply one’s probabilistic resources in order to increase the chances of the existence of finely tuned universes and then appeal to an observer self-selection effect to explain why we shouldn’t be surprised to observe that our universe is fine-tuned. Here is my objection:
In order to rescue the alternative of chance, its proponents have therefore been forced to adopt the hypothesis that there exists a sort of World Ensemble or multiverse of randomly ordered universes of which our universe is but a part. Now comes the key move: since observers can exist only in finely tuned worlds, of course we observe our universe to be fine-tuned! So this explanation of fine-tuning relies on (i) the existence of a specific type of World Ensemble and (ii) an observer self-selection effect.
Now this explanation, wholly apart from objections to (i), faces a very formidable objection to (ii), namely, the Boltzmann Brain problem. In order to be observable the entire universe need not be fine-tuned for our existence. Indeed, it is vastly more probable that a random fluctuation of mass-energy would yield a universe dominated by Boltzmann Brain observers than one dominated by ordinary observers like ourselves. In other words, the observer self-selection effect is explanatorily vacuous. As Robin Collins has noted, what needs to be explained is not just intelligent life, but embodied, interactive, intelligent agents like ourselves. Appeal to an observer self-selection effect accomplishes nothing because there’s no reason whatever to think that most observable worlds or the most probable observable worlds are worlds in which that kind of observer exists. Indeed, the opposite appears to be true: most observable worlds will be Boltzmann Brain worlds.
Since we presumably are not Boltzmann Brains, that fact strongly disconfirms a naturalistic World Ensemble or multiverse hypothesis.
Notice that I am willing, for the sake of argument, to concede the highly controversial assumption (i) that a very specific type of World Ensemble (or multiverse) exists. Rather I go after (ii), the observer self-selection effect. The fact that I bypass objections to (i) is a measure of just how vulnerable (ii) is. Multiverse defenders of the chance hypothesis gratuitously assume that either most of the observable universes in the World Ensemble are finely tuned or else that finely tuned universes, though fewer, are for some reason more probable than observable worlds that are not finely tuned. As I noted, not only is there no positive reason to think that either of these assumptions is true, but they appear to be false. This is hugely problematic for Carroll, who appeals to the multiverse to explain fine-tuning as due to chance.
Again, the ball has been put squarely in his court: in order to defend (ii), he needs to show that most observable worlds are finely tuned or that for some reason finely tuned observable worlds are more probable worlds. What will he say?
Carroll’s Response to the Boltzmann Brain Problem
Well, look at his response in his opening speech. He chooses to pass over the Boltzmann Brain problem in the context of the cosmological argument, commenting, “Finally, he makes a big deal about Boltzmann Brains. I’m going to talk about that a little bit later.” When we come to his discussion in the context of the teleological argument, the Boltzmann Brain problem comes up in his defense of the multiverse hypothesis (his fourth objection to the teleological argument). Here is what he has to say:
Now Dr. Craig makes a lot about the Boltzmann Brain problem, the problem that in the multiverse we could just be random fluctuations rather than growing in the aftermath of a hot big bang. This is a significant misunderstanding of how the multiverse works. The multiverse doesn’t say that everything that can possibly happen happens with equal probability. It says that there’s a definite history of the multiverse and you can make predictions. Different multiverse models will have different ratios of ordinary observers to random observers. That’s a good thing. That helps us distinguish between viable models of the multiverse and non-viable models, and there are plenty of viable models where the Boltzmann Brain, or random fluctuations, do not dominate. Furthermore, just as a little preview of coming attractions, I’m trying to write a paper (when I’m not debating about God and cosmology; I’m a physicist). I’m currently working on a paper that says, actually, Boltzmann Brains (random fluctuations) occur much, much less frequently than we previously believed. It comes down to a better understanding of quantum fluctuations. There’s a caricature of theism that says theism is an excuse to stop thinking. You say, “Oh, there’s a problem, I don’t need to solve it because God will solve it for me.” That’s clearly false because many theists think very carefully and very rigorously about many problems. But sometimes there’s an element of truth to it. This is an example. You’re faced with the Boltzmann Brain problem and you go, “I get out of that by saying that God created a single universe.” That might have stopped you from thinking about the physics in a deeper way and discovering interesting facts like this.
Does this reply answer the challenges I posed in my opening speech? Has Carroll returned service? Clearly not! His reply amounts to mere hand-waving, with a promissory note attached. He just asserts that there are viable multiverse models in which Boltzmann Brains do not dominate observable worlds. No specifics or documentation is given. Notice that he makes no attempt to show that Boltzmann Brains would not dominate in observable worlds according to his model, which is his best shot at solving the low entropy problem. Instead, he appeals to unspecified models which are not even described, much less shown to be plausible.
I’m not sure of the contents of his unpublished paper, but as an illustration of the problems Carroll faces, consider one suggestion he has made in his blog. There Carroll advises that we “need to destroy the universe (or at least the state it’s currently in) to save ourselves from the invasion of the Boltzmann Brains.” (That’s pretty strong medicine!) So he speculates that maybe our universe is hung up in a false vacuum state, and before Boltzmann Brains can form, the universe will quantum tunnel to a zero- or negative-energy vacuum state in which Boltzmann Brains cannot arise.
Unfortunately, Carroll requires a good deal of ad hoc physics to ensure that the lower energy vacuum state will be a zero- or negative-energy vacuum. He gives three options: we must either (i) say that the consensus about the mass of the top quark is wrong (which he admits is unlikely) or (ii) appeal to a new physics (an alternative he rejects when it comes to his own calculations) or (iii) adopt a minority view about the probability measure governing worlds. All of these are ad hoc and implausible moves.
Moreover, it’s not enough to destroy our universe. To avoid the Boltzmann Brain problem Carroll has to say that all or most of the baby universes are hung up in this higher vacuum state. Given their independence of one another, that is highly improbable. Even if ours is, why think that the others are? As Robin Collins pointed out in discussions prior to the debate, “one would have to hypothesize a special multiverse that varied the other constants, but was such that it did not vary the [relevant] masses from universe to universe. That is very ad hoc and implausibl.”
Rather than destroy the universe, isn’t a simpler way of avoiding the Boltzmann Brain problem to say that the universe began relatively recently in a low entropy state? This is not the God-of-the-gaps caricature that Carroll criticizes. One avoids the Boltzmann Brain problem by holding that the universe began to exist relatively recently in an initial low entropy condition—precisely what most cosmologists think. God doesn’t even enter the picture at this point.
Counter-Response to Carroll
In my rebuttal speech, in reviewing the kalam cosmological argument, I pointed out Carroll’s failure to defend his model against the Boltzmann Brain objection:
I also, secondly, pointed out that there is a Boltzmann Brain problem with respect to Dr. Carroll’s model. It seems to me that he just didn’t respond to the point that I was making, namely, that since every baby universe grows into a de Sitter space, there will be vastly, vastly more of these Boltzmann Brains in the long run than there will be ordinary observers. So what Dr. Carroll would need to do is to justify some non-standard measure of probability that would make ordinary observers more probable than Boltzmann Brains. But he admits that he cannot do it.
Remember that his model represents his best effort to come to grips with the initial low entropy condition of the universe; that is its entire raison d’être. But now he declines to defend it.
When it comes to the teleological argument, I sought to advance the discussion by exposing a further weakness of the multiverse explanation regarding Boltzmann Brains:
. . . he confronts the Boltzmann Brain problem once again. Even if Dr. Carroll could show that ordinary observers predominate in life-permitting worlds, what about all of those worlds that are not life-permitting because the other constants and quantities are not finely-tuned? In such worlds, which vastly outnumber finely-tuned worlds, there will be no ordinary observers, and yet there will be untold numbers of Boltzmann Brains produced by thermal fluctuations. So the entire multiverse will be dominated by universes having vastly more Boltzmann Brains than ordinary observers like us. Therefore, the data on the multiverse hypothesis is incomprehensibly improbable. The evidence is strongly disconfirmatory of the World Ensemble hypothesis.
This is a really devastating objection, which I owe, as noted, to discussions with Robin Collins. The point is really quite obvious. Suppose Carroll could somehow show that in finely tuned universes (like ours) ordinary observers will dominate, perhaps due to the ad hoc physics appealed to above. That still wouldn’t rescue the multiverse hypothesis from the Boltzmann Brain problem! For what about all the worlds that are not finely tuned? Because these universes have no ordinary observers at all and are incomprehensibly more plenteous than finely tuned universes, the multiverse as a whole will be dominated by Boltzmann Brains. Carroll’s confident assurances that there are viable multiverse models that escape the Boltzmann Brain problem do nothing to show how to solve this problem.
So when we come to Carroll’s rebuttal speech, we’re still waiting to see how he will meet the challenges I posed both to his specific model and to the World Ensemble hypothesis in general. With respect to the kalam cosmological argument, again he backs away from a defense of his model against the Boltzmann Brain objection: “On a more specific level we talked about my model. Again, I’m not trying to defend my model; I’m the first one to say that it has problems.” He claims that its problems are not the ones I mention; but we’re still waiting to hear how it solves the Boltzmann Brain problem. If his model can’t solve that problem, then it loses its raison d’être, since it fails as a plausible explanation of the initial low entropy condition of the universe.
With respect to the teleological argument, Carroll again resorts to handwaving rather than give us the specifics:
The main argument Dr. Craig has against this is the Boltzmann Brain argument. Again, I gave you why that’s not a good argument and it seems to have been ignored in Dr. Craig’s last speech; namely that the multiverse does not predict that everything happens. It predicts certain things happen with certain frequencies. So what you do as a working cosmologist is check that your universe is not dominated by Boltzmann Brains. Are there multiverse models that pass that test? Yes, there are. It is easily avoided. Then he brings up this weird argument, he says, “There could be the regions where ordinary observers could not exist because the parameters are not right but Boltzmann Brains could exist.” But the whole point is that Boltzmann Brains are forms of life. Boltzmann Brains can only exist where life is possible. So what one does in cosmology is look at the multiverse regions where life is possible and counts the number of Boltzmann Brains versus the number of ordinary observers—there are models that pass the test.
Carroll is quite right that I did not respond to his point that multiverse hypotheses do not predict that everything will happen. That is simply because I have never labored under that misimpression and it plays no role in my argument. Quite the contrary, different universes in the World Ensemble will be more or less conducive to embodied, conscious observers based on the degree of fine-tuning which their fundamental constants and quantities exhibit. And the objection is that no explanation has been given why the World Ensemble, considered as a whole, will be not be dominated by universes featuring Boltzmann Brains rather than respectable observers like us. You can avoid this problem by adopting a special measure of probability favoring those relatively rare universes with ordinary observers, but then you owe us some explanation for this move.
Lest we lose the forest for the trees, recall that what is at stake here is the appeal to an observer self-selection effect in order to justify why we observe a low-entropy, finely tuned universe around us. For this appeal to work, the chances have to be such that a random observer will likely be an ordinary observer rather than a Boltzmann Brain. But Carroll has never shown that. Carroll’s point that Boltzmann Brains are forms of life only underlines Robin Collins’ claim that what is at stake in the fine-tuning argument is not the frequency of life-permitting worlds but worlds permitting embodied, conscious agents. Some explanation is needed why this latter sort of world exists, and here appeal a self-selection effect is, as I explained, vacuous.
So it seems to me that I am quite justified when, in my closing statement, with reference to the Boltzmann Brain problem, I complain,
I don’t think that Dr. Carroll has really come to grips with this, quite honestly. There are at least two reasons why Boltzmann Brains would dominate. First, because on his multiverse model, in the long run every baby universe becomes a de Sitter space and will become dominated with Boltzmann Brains. Secondly, in all of the other worlds that are not fine-tuned, there just aren’t any ordinary observers; but there will be thermal fluctuations that will produce Boltzmann Brains. So he is the one who has to justify some non-standard measure of probability in order to explain why ordinary observers like us should exist rather than Boltzmann Brains.
That sums it up rather nicely. Perhaps Carroll can answer these problems; but in the context of this debate he has not done so. He has resorted to handwaving rather than evidence and has just asked us to trust him.
In his final speech Carroll says with respect to his model, which he declined to defend, “[Craig] says that there should be more Boltzmann Brains than ordinary observers. I again explain why that’s not true because it’s a model dependent statement. In this particular model it turns out to be easier to make a universe than to make a brain. That’s a selling point of the model.” Really? Then that ought to have been defended. In his book Carroll says at most that it’s “plausible.” Even that is being optimistic. Carroll needs to respond specifically and substantively to explain how his model will avert the objections I lodged against it and why his model will, in fact, yield more ordinary observers than Boltzmann Brains.
Obviously much more was said in the debate concerning the evidence of thermodynamics for the beginning of the universe and concerning the hypothesis of chance as an explanation of fine-tuning. But I hope that by focusing on the Boltzmann Brain objection, I have, at least, illustrated how to assess the argumentation in a debate. You isolate an issue and then pursue that thread through the sequence of speeches (and through the answers to questions afterwards), seeing how the response and counter-response proceeds.
When we do that exercise with respect to the Boltzmann Brain objection, it becomes apparent how weak Carroll’s response is. With respect to the thermodynamic evidence for the second premiss of the kalam cosmological argument, Carroll’s attempt to avert the beginning of the universe by means of his time-reversal model fails to provide a convincing answer to the Boltzmann Brain problem. There are two reasons for this. First, since de Sitter space spawns Boltzmann Brains as well as baby universes and each baby universe becomes a de Sitter space itself, all the ordinary observers in any universe will eventually die out and Boltzmann Brains will dominate in the long run. Second, even if in finely tuned universes like ours, ordinary observers dominate, the fact remains that in the worlds which are not fine-tuned, Boltzmann Brains will vastly dominate, so that the World Ensemble will be dominated by universes whose observers, if any, are Boltzmann Brains. This result strongly disconfirms Carroll’s model as an attempt to undercut the evidence of thermodynamics for the beginning of the universe.
With respect to the attempt to rescue the alternative of chance as an explanation of the fine-tuning of the universe by appeal to the World Ensemble hypothesis, Carroll merely asks us to trust him that there are viable models that can avoid the Boltzmann Brain problem. Personally, I have no confidence in Carroll’s assurances because he said something similar with respect to viable models of a beginningless universe, and when one gets down to specifics all of those models turn out to be very problematic and implausible. But never mind; the more important point is that in a debate context assertion is no substitute for argument. Carroll needs to gives us the specific evidence for his claims if his assertions are to have any weight.
This post and other resources are available on Dr. William Lane Craig's website: www.reasonablefaith.org
 Carroll’s accusation that mine is God-of-the-gaps reasoning is clearly wrong, since I appeal to the thermodynamic evidence merely to show that the universe had a beginning. Whether that beginning has a cause of any sort is an independent issue, discussed in QoW #368. Carroll thus fundamentally misunderstands me when he alleges that I am offering theism as an alternative cosmological model.
 Responding to Carroll’s three alternatives, Collins wrote,
“The solutions Carroll proposes . . . to BBs arising in the future of our universe seems to present two problems. In the Higgs and quark mass solutions, there needs to be a pretty severe fine-tuning of these masses as Carroll notes. . . . This fine-tuning. . . cannot be explained by a multiverse, since in a multiverse the vast proportion of universes will be ones in which the mass is different, and hence an indefinite number of BBs will arise in these universes’ future, and hence BBs will again dominate over the entire multiverse.
“Further, adopting this line will present a major problem to a multiverse explanation of the other constants: one would have to hypothesize a special multiverse that varied the other constants, but was such that it did not vary these masses from universe to universe. That is very ad hoc and implausible.
“As for the measure solution, the question here is whether the needed measure is contingent or a priori. If contingent, then the question arises as to whether it varies from universe to universe, like other parameters. If it does, the problem is the same as above for the Higgs and quark masses. If not, we would need a very good a priori justification that this was the right measure.”