Preaching to the Unconverted

And now for something somewhat different. After I posted my article on “Does the Universe Need God?“, there were a few responses at the Intelligent Design blog Uncommon Descent, including a list of questions by Vincent Torley. Vincent then went the extra mile by inviting me to write a guest post for UD. Not my usual stomping grounds, but I ultimately agreed, precisely for that reason.

Here’s the post, which I’m cross-posting below. This might be controversial, as a lot of people on my side of things will say that there’s little point in engaging with people on the other side. And admittedly, this is a subject where feelings can be pretty entrenched. But you never know — not everyone has their mind made up on every issue, and it’s good to try to explain yourself to unsympathetic audiences on occasion. That’s all I tried to do here — to explain how I think about these things, not necessarily to pick a fight or even persuade any skeptics. I tried pretty hard to be as clear and unpretentious as I can be. (Success is for you to decide.) In a world of shouting and diatribe, I remain optimistic that real communication can occasionally occur! We’ll see how it goes.


I wanted to thank Vincent Torley and Denyse O’Leary for the opportunity to write a guest blog post, and apologize for how long it’s taken me to do so. I’ve written an article for the forthcoming Blackwell Companion to Science and Christianity, entitled Does the Universe Need God?, in which I argued that the answer is “no.” Vincent posed a list of questions in response. After thinking about it, I decided that my answers would be more clear if I simply wrote a coherent argument, rather than addressing the questions individually.

My goal is to try to explain my own thinking to an audience that is not predisposed to agree. We can roughly break people up into two groups: naturalists such as myself, who think that the best explanation we have for the universe involves physical quantities obeying laws of Nature and nothing else; and those who believe that a better explanation can be found by invoking a powerful being/designer/creator/God. (For the sake of simplicity I’m going to use “God” to refer to this notion, but feel free to substitute the more accurate description of your choice.) Obviously there are many nuances that are being passed over by this simple distinction, but hopefully it will suffice for this moment.

The dispute between these two camps isn’t one where people often change their minds at the drop of an argument. Minds do change, in either direction — but typically after extended periods of reflection, not suddenly in response to a single killer blog post. So persuasion is not my goal here; only explanation. I’ve succeeded if an open-minded person who disagrees with me reads the post and still disagrees, but at least understands why I hold my positions. (After giving an earlier talk, one of the theologians in the audience told me that I had persuaded him — not that God didn’t exist, but that the argument from design wasn’t the way to get to Him. That sort of real-time response is more than one can generally hope for.)

What I want to do is to elaborate on some crucial aspects of how science is done that bear directly on the issues raised by my article and some of the responses to it that I’ve seen. In particular, I want to talk about simplicity, laws, openness, explanation, and clarity. This isn’t supposed to be a comprehensive treatise on the philosophy of science, nor is it especially rigorous, or anything really new — just some thoughts on issues relevant to this conversation.

I will be taking one thing for granted: that what we’re interested in doing here is science. There are many kinds of consideration that may lead people to theism or atheism that have nothing whatsoever to do with science; likewise, one may believe that there are ways of understanding the natural world that go beyond the methods of science. I have nothing to say about that right now; that’s a higher-level discussion. I’m just going to presume that we all agree that we’re trying to be the best scientists we can possibly be, and ask what that means.

With all that throat-clearing out of the way, here’s what I have to say about these five issues.


Science tries to capture the world in the simplest possible description. We are fortunate that such an endeavor is sensible, in that the world we observe exhibits various regularities. If the contents and behavior of the world were completely different from point to point and moment to moment, science would be impossible. But the regularities of the world offer a tremendous simplification of description, making science possible. We don’t need to talk separately about the charge of this electron, and the charge of that electron; all electrons have the same charge.

Simplicity can be quantified by the concept of Kolmogorov complexity — roughly, the length of the shortest possible complete description of a system. It takes longer to specify some particular list of 1,000 random numbers than it does to specify “the integers from 1 to one million,” even though the latter contains more elements. The list of integers therefore has a lower Kolmogorov complexity, and we say that it’s simpler. Scientists are trying to come up with the simplest description of nature that accounts for all the data.

Note that a theory that invokes God (or any other extra-physical categories) is, all else being equal, less simple than a theory that does not. “God + the natural world” is less simple than “the natural world.” This doesn’t mean that the idea of God is automatically wrong; only that it starts out at a disadvantage as far as simplicity is concerned. A conscientious scientist could nevertheless be led to the conclusion that God plays a role in the best possible scientific description of the world. For example, it could (in some hypothetical world) turn out to be impossible to fit the data without invoking God. As Einstein put it: “It can scarcely be denied that the supreme goal of all theory is to make the irreducible basic elements as simple and as few as possible without having to surrender the adequate representation of a single datum of experience.” Alternatively, you could imagine deriving all of the physical laws from the simpler assumption that God exists. While these strategies are conceivable, in practice I don’t think they work, as should become clear.


A “law of nature” is simply a regularity we observe in the universe. All electrons have the same charge; energy and momentum are conserved in particle interactions. A law doesn’t necessarily have to be absolute or deterministic; the Born rule of quantum mechanics states that the probability of obtaining a certain observational result is the square of the amplitude of the corresponding branch of the wave function. A law is simply a pattern we observe in nature.

As far as science is concerned, it makes no difference whether we refer to these regularities as “laws” or “patterns” or anything else. It also doesn’t matter whether we think of them as “fundamental and irreducible features of the cosmos.” They simply are; science looks for them, and finds them. Vincent asks “How can rules exist in the absence of a mind?” That is simply not a question that science is concerned with. Science wants to know how we can boil the behavior of nature down to the simplest possible rules. You might want more than that; but then you’re not doing science. He also asks why we should believe that the rules should continue to hold tomorrow, simply because they have held in the past. Again, that’s what science does. Imagining that the same basic laws will continue to hold provides a simpler fit to the data we have than imagining (for no good reason) that they will change. If you are personally unsatisfied with that attitude, that’s fine; but your dissatisfaction is not a scientific matter.


This is probably the most important point I have to make, and follows directly on the issue of “laws” just addressed. There is a way of trying to understand the world that might roughly be called “scholastic,” which sits down and tries to reason about how the world should be. The great success of science over the last five hundred years has been made possible by throwing out that kind of thinking in favor of a different model. Namely: we think of every possible way the world could be, and then we go out and look at the world to see which is the simplest description that fits the data. Science insists that we be open to all possibilities, and let the data decide which is true.

Suppose that you are convinced that laws of nature could not exist without a guiding intelligence that formulated them and sustains them. That’s fine for you, but it’s a deeply unscientific attitude. The scientific attitude is: “We observe that there are regularities in nature. We might imagine that they are formulated and sustained by a guiding intelligence, or that they simply exist on their own. Let’s go collect data to determine which idea is a more parsimonious fit to reality.”

The primary sin a scientist can commit is to decide ahead of time that the universe must behave in certain ways. We can certainly have intuitions about what kind of behavior “makes sense” to us as scientists — theorists are guided by their intuition all the time. But the use of that intuition is to help us develop hypotheses, not to decide which hypothesis is correct. Only confrontation with data can do that.


Science has a complicated relationship with “Why?” questions. Sometimes it provides direct answers: Why do all electrons have the same charge? Because they are all excitations of a single underlying quantum field. But sometimes it does not: Why is there a quantum field with the properties of electrons? Well, that’s just the way it is. Which questions have sensible answers is dependent on context, and can even change as we learn new things about the universe. To Kepler, understanding why exactly five planets orbit the Sun was a question of paramount importance. These days we think of the number of planets (eight, according to the International Astronomical Union) as something of an accident.

The point, once again, is that we can’t decide ahead of time what kinds of explanations science is going to provide for us. Science looks for the simplest possible description of the world. It might be that we will eventually understand the inner workings of nature so well that we will be able to answer every conceivable “Why?” question — we will ultimately see that things simply could not have been any other way. But it is also perfectly possible that the best possible description of the world involves some number of brute facts that have no deeper explanation. This is an issue that will ultimately be decided by the conventional progress of science, not by a priori demands that the universe must explain itself to anyone’s individual satisfaction.


The final point I wanted to make involves the clarity of scientific hypotheses. Perhaps “unambiguity” would have been a more precise word, but it is so ugly I couldn’t bring myself to use it.

The point is that a respectable scientific theory should be formulated in terms that are so unambiguously clear that any two people, both of whom understand the theory and have the technical competence to elucidate its consequences, will always come to the same conclusion about what the theory says. This is why the best theories we have are very often cast in the form of mathematics; the rules for manipulating equations are absolutely free of ambiguity. You tell me the initial conditions of some classical mechanical system, as well as the Hamiltonian, and I will come up with the same predictions for its future evolution as absolutely anyone else wit the same information.

Earlier I mentioned that the God hypothesis could actually be simpler than a purely naturalistic theory, if one could use the idea of God to derive the observed laws of nature (or at least some other features of the universe). This isn’t idle speculation, of course; many people have taken this road. The fundamental problem, however, is that the idea of God is utterly unclear and ambiguous, as far as conventional scientific thinking is concerned.

One might object: God is simply the most perfect being conceivable, and what could be more unambiguous than that? (One possible response, not the only one.) That sounds like a clear statement, but it’s not in any sense a clear scientific theory. For that, there would have to be a set of unambiguous rules that let you go from “the most perfect being” to the laws of nature that we see around us. As I argued in my paper, this is very far from what we actually have. It is sometimes argued, for example, that God explains the small value of the vacuum energy (cosmological constant), because without that fine-tuning life would be impossible. But why does God choose this particular value? Actually it could be quite a bit larger and life would still be very possible. Why are there 100 billion galaxies in addition to the one we live in? Why are there three generations of elementary particles, when life is only constructed from the first one? Why was the entropy of the early universe enormously smaller than it needed to be to support life?

Obviously these are perfectly good questions for naturalistic theories as well as for God. The problem is that we can imagine coming up with naturalistic theories that do provide clear answers, while it’s very hard to see how God could ever do that. The problem is simple: God isn’t expressed in the form of equations. There is no clear and unambiguous map from God to a particular set of laws of physics, or a particular configuration of the universe. If there were, we would be using that map to make predictions. What does God have to say about supersymmetry, or the mass of the Higgs boson, or the amplitude of gravitational-wave perturbations of the cosmic microwave background? If we claim that God “explains” the known laws of physics, the same method of explanation should work for the unknown laws. It’s not going to happen.

It’s not clear to me that anyone who believes in God should actually want it to happen. There is a very strong tension between what scientists look for in a theory — clear and unambiguous connections between premises and predictions — and the way that religious believers typically conceive of God, as a conscious being that is irreducibly free to make choices. Does anyone really want to reduce God to a simple set of rules that can be manipulated by anyone to make clear predictions, like we can in theories of modern physics? If not, God will always remain as a theoretical option of last resort — something to be invoked only after we are absolutely convinced that no possible naturalist option can explain the universe we see.


Obviously these very simple points don’t come anywhere near addressing all the possible issues in this area. In particular, I haven’t made any real attempt to argue that a purely naturalistic explanation actually is a better fit to the observed universe than God or similar ideas. Instead I’ve just tried to explain the mindset of someone like me who does end up coming to that conclusion. In my paper I’ve tried to lay out why invoking God doesn’t seem to provide an especially promising explanation of the world around us. Others may disagree, but I hope this has made things more clear.

This entry was posted in Philosophy, Religion. Bookmark the permalink.

83 Responses to Preaching to the Unconverted

  1. DaveH says:

    @Tyro #24,

    (From Sean,) We can say that we aren’t Boltzmann Brains because the rest of the universe isn’t in thermal equilibrium. If we were typical Boltmann Brains we shouldn’t see much.

    Both solipsism and the notion that I am a Boltzmann Brain are cognitively unstable. At best, the solipsist is arguing with himself.

    It’s possible that the consistent narrative we have regarding the empirical world (13.72bn yr old universe consistent with The Standard Model of physics, quantum mechanics and GR; cosmological evolution of solar system and planets; 4.5bn yr old Earth; Darwinian evolution etc etc) could be the work of a deceptive demon, but it’s not an idea we should take seriously.

    So I don’t think the commenter is onto something. And solipsism doesn’t get him closer to justifying his belief in God.

  2. Jim Cross says:

    The belief in God and the belief in the scientific method actually seem to have a great deal in common. I find it somewhat curious that each side of the argument seems to want to draw such sharp distinctions.

    For one thing, both sides start with belief and it is, in fact, belief about the almost identical thing – the inherent orderliness of the world. Without orderliness and predictability, the entire scientific effort would be impossible. The scientific enterprise is an effort to explain the world through discovery of its inherent order. The fact of its order is unquestioned. Religious belief begins also with orderliness and merely adds a Deity as the Original Cause. Proponents of the scientific method may argue that adding God into the explanation is unnecessary; however, the scientific method is left with an inexplicable orderliness of the universe at its foundation.

    When we think as mathematicians we have no problem in understanding that there is no highest number. We have created the concept of infinity to express this. Yet when we come to search for causation as scientists we seem unable to accept a similar concept – that there may be an endless (literally) chain of causation. Some may some choose to conceptualize this endless chain of causation that gives order to the world as God.

  3. AnotherSean says:

    Its certainly fair to say God is ambigous and undeffined. Whether this is a matter of logical necessity or rehtorical convenience is a matter of opinion. Either way, I think it means the concept of God cannot be used as a logical symbol or natural explanation.

  4. bob says:

    God is not a necessary progenitor of, or component of the universe, considered from the point of view of astronomy. But viewed from the perspective of what it means to be human, God is necessary for our emotional well being. I think that, as our neocortex emerged, there was not the recognition of “self”. That inner voice that we recognize as the avatar of our own consciousness, may not have been so in our early developement. The voice in our heads may have been experienced as coming from outside ourselves, thereby creating a meme for God or gods early on in our evolution. I feel a sense of wonderment when I think about “God, the Universe and Everything”, which maybe is the whole point.

  5. ohwilleke says:

    Even if the universe doesn’t need a god, humans may need a religion, theistic or otherwise.

  6. bob says:

    For Sean. “the Anthropic Cosmological Principle” and “Godel, Escher, Bach”.

  7. bob says:

    ohwilleke: I think I just said that.

  8. Jessie Desmond says:

    For a god to be instantaneously aware/interactive of the universe, and consistent with observation, he must be entangled with it. Observation causes entanglement collapse. The only consistent compromise is a very large number of very small incremental observations. Anybody claiming to know god is then either wrong, or correct and kills god instantaneously throughout the universe.

    Sign a pact with the Devil and all the pressure is off. Locality rules!

  9. vel says:

    “: God is simply the most perfect being conceivable, and what could be more unambiguous than that?” Well, who is doing the conceiving? Why is this “prefect” being, given vastly different traits by each religion of what it “wants”?

    and I have no need of “god” to make me emotionally well. I get along quite well contemplating the universe without having to have a bogeyman to make me feel like a special snowflake.

  10. Tyro says:

    “: God is simply the most perfect being conceivable, and what could be more unambiguous than that?”

    If there is such a thing as perfection and it’s singular rather than being a broad landscape with many peaks, surely science can investigate it. In fact, only science can do so. As a part of this perfection, wouldn’t this also mean that there would be a singular, perfect moral code which this being would possess/embody and which we could derive independently? Well, would anyone care to step up and try this task? Sam Harris just wrote a book about how morality can be investigated scientifically and he was heavily criticized for this – let’s see some theists making this perfection argument stand up and defend him.

    Okay, let’s set aside morality. What other qualities are a part of this perfection? Skin tone? Favourite ice cream? Is it a tenor or baritone? What is the perfect gender or is perfection gender-less (and does this mean that sex is not a part of perfection, no matter how pleasurable it feels)? Please sketch out some of the attributes we can infer based on this premise.

  11. spyder says:

    I come at this from the study of religions, and this is my short take. Charles Long wrote a book a long time ago called ALPHA: Myths of Creation, in which he describes around 40 different creation myths from around the world. The book makes it easy to pick and choose which one fits the ID model best. It also allows the reader to understand that there certainly is no one god or goddess or other supernatural beingness, but hundreds of possible ones. Much like Herzog in his new film Cave of Forgotten Dreams, the capacity for human creative thinking far surpasses those of his gods.

  12. ttsuchi says:

    “One commenter on UD just said that God was described as simple, case closed. Instead of undermining complexity, that just seems to show that this description of God is wrong.”

    I think one way to get the point across is to phrase the “simplicity” from the perspective of making falsifiable predictions. “Simple” doesn’t refer to the complexity of the explanation itself; it refers to the number of free parameters of the theory. (Simple explanation is good in terms of *understanding* the theory, but that’s a different matter…) Theories with smaller number of free parameters can make larger number of falsifiable predictions, and that’s why “simple” theories are more valued in science. Omnipotence and solipsism are terrible because they explain everything (not only what does happen, but also what could happen) too well. Those theories have infinite explanatory power and zero predictive power…

  13. Unclellama says:

    @27 Jim Cross: the problem here is the ambiguity of the word ‘belief’. It covers two wildly different notions: information about the world filtered through our senses (I believe that I am typing this message, because that’s what my senses and knowledge of the message-typing situation are telling me; I believe that the Universe follows certain laws because it keeps on acting as though it does), and faith.

    While it is quite possible that the existence of some kind of creator could be argued for along scientific lines (so far none of the arguments I have read are in any way convincing, but that’s not necessarily God’s fault), all the religions I know of are 100% reliant on that other kind of belief. And THAT kind of belief is a long way away from science. My ‘belief’ in the scientific method is simply that I have noticed how good a job it has done so far. That has nothing to do with faith, unless you bend that word almost to breaking point.

  14. SLC says:

    I think that Prof. Carroll is missing the biggest problem with accepting god as an explanation. It is that a theory based on god is unbounded, since god is all powerful and can therefore do anything. Thus,it is impossible to make a prediction based on god did it.

  15. Lord says:

    Science is great for doing science. Where it has problems is when people try to use it to go beyond it. Science is an explanation of the natural world. It can’t say anything about the supernatural world without making it part of the natural world. Science considers all manner of ideas that have no evidence in the natural world; it could hardly progress otherwise. Do wormholes exist? They aren’t simple. They aren’t known. There is no need for them. Yet they don’t dismiss them on those grounds. They attempt to determine whether they could exist or do exist and any such attempt must center around possible data or a model that data could be gathered that could confirm or refute it. We may have no such data but would we even have such data or recognize it as such without such a model? Science will not be able to say they don’t exist without a model of them and even then only say that that model of wormholes cannot exist, not that there might not be some other model that might. On the other hand, it might discover or create them with such a model. Does God exist is a meaningless question to a scientist without a model that could be tested and useless as an explanation without such a model, but there will always be areas beyond the natural world unamenable to science, perhaps only currently but perhaps eternally.

  16. Gunner says:

    You should have just told them to read your post “Physics and the Immortality of the Soul”, it could have been quite possibly the most logical thing they have ever read.

  17. Charon says:

    To Kepler, understanding why exactly five planets orbit the Sun was a question of paramount importance.


    Kepler’s system was heliocentric, as you say, so… there were six known planets. The five Platonic solids were supposed to separate the orbits of the planets.

    Try as I may, I generally agree with everything Sean writes, so when I don’t, I have to point it out. Even if it’s incredibly nitpicky 😉

  18. Charon says:


    there will always be areas beyond the natural world unamenable to science

    This is a common claim, yet not one I’ve ever seen actually explained. This is a really vexing problem for those who claim there is a “supernatural” realm that can’t be studied physically, but that also interacts with the natural realm (otherwise, it clearly has no impact on our lives, and is equivalent to not existing). This is why Descartes’ separation of the body and the soul then proceeded to confuse everyone for hundreds of years, leading to utter nonsense like the Occasionalism of Malebranche.

    Somehow science can study the extremely subtle effects that dark matter and neutrinos have on ordinary atoms, but it is claimed, for no apparent reason, that science can’t study the much stronger effects of a supposed spiritual realm? This is called special pleading.

  19. Charon says:

    @rob j

    I have experienced things I cannot deny, with proof that cannot be discounted

    And this is what Sean means when he talks about that being fine, but the antithesis of science. In science, there is nothing you can’t deny, and nothing that can’t be discounted, given sufficient evidence to the contrary. Even things we hold very dear, like causality and locality (see Bell’s theorem). Bayes’ theorem provides a fun little example (okay, I have a weird idea of fun): if you set the probability p=0 or p=1 exactly, then it can never change! Regardless of the amount of new data collected! So there’s a tip – if you ever set p=0 or p=1, as you’ve just done, you are no longer doing science. What you are doing is utterly incompatible with science.

    That might be fine for you. But you’re not allowed to claim it is compatible with science when it’s clearly not.

  20. math says:

    @ ALL
    I think its fairly obvious that when one talks of the soul, the one is really referring to the recorded knowledge in one’s brain. It should be obvious to even the most closed minded physicist that we can discuss whether the physics of the universe supports the continued recordation or recurrence of the quantum states associated with that knowledge. However, where physics breaks down, and where physics becomes unphysical is the question of “why?”. To date the only answer has been, “why not?” Physicists may attempt to show that one can use an assumption of non-agency and achieve the same results, but they can not prove such a thing…no more than they can prove that there is a precise classical microstate that describes all of the universe at any instant. The result is that we must leave these questions to philosophy. “Proving a negative” might be extraordinarily difficult, but it is hypocritical to accept the need to do so in so many cases and yet say it is unnecessary in others.

  21. Jim Cross says:

    #38 Unclellama

    Quite a bit of modern physics (string theory? multiverse?) is probably as untestable as a God theory and there is a quite a bit of “faith” that the world is describable mathematically. What if it ultimately isn’t? Or that the order is so complex it cannot be reduced to a simpler description than the universe itself?

    I am not arguing for a God but just saying the scientific and religious viewpoints both begin with a fundamental belief in the orderliness of the universe and the differences between them are not all that great when we are looking for Theories of Everything. If we are simply talking about what works or is useful, I would rather rely on science to build a bridge rather than faith.

  22. Lord says:

    Do wormholes exist? It is really a vexing problem for those who believe they may but cannot be studied physically without a specific model but that they also interact with the natural realm. Now if you find one to study or a model that makes testable predictions, it is a different story, but until then wormholes are as supernatural as God.

  23. DaveH says:

    @Jim #46,

    there is a quite a bit of “faith” that the world is describable mathematically. What if it ultimately isn’t?

    We already know it is. What if all of it isn’t? Then, we won’t be able to describe it mathematically. It’s an empirical question.

    the differences between them are not all that great

    Yes they are. One method actually looks at the world.

  24. DaveH says:

    @Lord #47

    wormholes are as supernatural as God

    No-one thinks wormholes are supernatural, except you.

  25. Personalized says:

    I love the debate. Personally, if I can see it, smell it, or touch it, its not real 😉