Saturday, August 19, 2006

The Anthropic Principle

Dom Eggert writes:
[T]he likelihood that life will arise in this universe does not change even if there are millions of non-life-supporting universes "out there" (assuming that universes are closed systems that don't interfere with each other). If this reasoning is correct then, from our perspective, it is no more rational to believe that there are infinitely many universes than to believe there is just one--it's merely an aesthetic preference.

This strikes me as mistaken. It's true that the existence of other universes doesn't affect the probability of this universe - "u42" let's say - being capable of supporting life. But that isn't what needs explaining. We merely need to show how there could be some universe or other that supports life. The locative fact that it's ours in particular can be got for free, by appeal to the Anthropic Principle.

Compare: "Of all the planets in the universe, how is it that we ended up on one of the few capable of supporting life? Isn't this monumentally unlikely?" This sounds like a silly question. It's not as if we might instead have been asking the question from the blistered surface of Mercury. Living - and hence being somewhere capable of supporting life - is a precondition for even asking the question. Finding ourselves alive on a lifeless planet is not a possibility that ever needed to be ruled out.

A more troubling question would be: "How is it that there are any life-supporting planets at all?" For clearly the total absence of life is a coherent alternative, so we need some explanation of why that didn't come to be. (Otherwise we must appeal to brute chance or coincidence, but that isn't much of an explanation!) And here the appeal to multiple universes might help. If there are zillions of universes, the chance that life exists somewhere or other suddenly looks a lot more likely.

Note that once it has been explained how life could plausibly exist somewhere, it's no great mystery how come it exists in our universe in particular. Maybe it's unlikely that u42 would contain life, but we don't really care about u42. For explanatory purposes, we care about "the universe we're in", de dicto not de re. And the probability that whatever universe we're in contains life is pretty well certain.

Again, the Anthropic Principle by itself cannot solve the whole problem. It cannot address the question of why we (or life-supporting universes) exist at all. For that we need the multiverse hypothesis. The Anthropic Principle can only help with the locative question: presupposing that there are life-supporting locations, why do we find ourselves in one of them rather than somewhere else? Combine the two and we get a relatively satisfying explanation, I guess. (More so than any alternative I've yet come across, anyway.)

So, contra Dom, I think we could rationally believe the multiverse hypothesis, as an inference to the best explanation. We simply need to be clear on what it serves to explain. It does not explain why u42 supports life. That's not something I feel any need to explain. Rather, what the multiverse hypothesis initially explains is how any actual universe could support life. Conjoined with the anthropic principle, it can then explain why we find ourselves in a life-supporting universe.

(See also: Why does the universe exist?)


  1. Richard,

    For a multiverse hypothesis to be explanatory, it must preferentially predict our universe. For example, a multiverse theory might show that our universe's physical constants are not independent, but are related in some particular way. If the multiverse theory isn't predictive, then it is not explanatory of our universe, it just appeals to our aesthetic sense. Much the way that interpretations of quantum mechanics aren't explanatory, but appeal to aesthetics.

  2. What do you mean by "preferentially predict"? The multiverse theory I have in mind is the one which hypothesizes that every possible universe is actualized. That would certainly explain why our universe exists. I'm not sure what the problem here is meant to be...

  3. Well, I don't think that such a theory explains our universe when it predicts any observation we make, come what may.

    There's a spectrum of models from the highly explanatory (e.g., a model that predicts exactly what we see) to the non-explanatory. The highly explanatory models are not controversial. I think confusion seeps in when we fail to perceive that some models are equivalent to non-explanations.

    When looking for explanations, we are looking for solutions that meet criteria like this:

    There is some set of facts and rules, X, such that X implies our observations.

    For example, an observed car crash is implied by the driver losing consciousness. Thus, we would say that loss of consciousness is one possible explanation for a car crash. It may not be the only explanation, but it is an explanatory theory.

    However, we would not be satisfied with an explanation of the form "X caused the car crash, where X implies the car crash." That would just be a restatement of the problem. No matter how much we learn about the crash, X is always infinitely tunable. X always reflects our current knowledge of the crash, but never predicts an actual observation (e.g., never predicts blood alcohol levels, etc.).

    Thus, I think the test of an explanation is its ability to predict observations. If it's not predictive at all, then we're saying that the explanation is infinitely tunable to match whatever we may discover. In other words, it's a restatement of the puzzle, not an answer to it.

    A multiverse theory of the type you describe is compatible with anything we might discover about our universe. It's also compatible with anything we might discover about how our universe might have been. So, it really just seems to restate what we already know about universes, without answering any questions at all. The infinite collection of universes is just a wildcard for all the possibilities consistent with what we already know. (The same is true of God, BTW.)

    In contrast, if we have a respectable multiverse model that, say, predicts a non-trivial relationship between the speed of light and Planck's constant in all universes, then we're really getting somewhere. Such a model preferentially predicts our universe in a testable way. Our universe would at least behave as if it were part of an ensemble of universes, even if those universes were a convenient fiction.

    I'll just mention that postdiction (predicting one subset of your data from the remaining subset) is about as good as prediction. However, the kind of model you cite cannot even do postdiction. Knowing that all possible universes exist, and knowing a subset of what we know about our universe, doesn't enable us to predict the remaining subset of facts we know about our universe.

  4. I don't think you've given me any reason to accept that "predict[ing] exactly what we see" is the sole criterion of a good explanation. For present purposes, I simply want to understand how it is that a life-supporting universe exists, given that most possible universes are presumably not life-supporting. One explanation, as offered in the main post, is that all possible universes (rather than just one special one) exist. If true, this fact would explain why life exists. Life exists because it is possible, and everything possible exists. Puzzle solved. That's an explanation. Surely. It doesn't matter whether it "preferentially predicts" the particular observations found in my locality -- that isn't what I'm trying to explain, so such demands would seem to miss the point.

  5. Life exists because it is possible, and everything possible exists.

    It might be ok to say the probable existence as a result of possibility, but neveer justified the certainty of existence. You seem to be too hasty so as to miss the basis.

    Also, I find what Doctor Logic said quite plausible. Multiverse can serve only as a placebo of our insatiable appetite for knowledge. Since it cannot be disproved nor proved, it is not a "Scientific Hypothesis".

  6. The multiverse is the biggest cop-out on causality responsible science since uncertainty killed purpose in nature.

    I've said it before, and I'll likely say it agian...

    Einstein should have killed Bohr and Heisenburg, instead.

  7. Richard,

    You say:

    I simply want to understand how it is that a life-supporting universe exists, given that most possible universes are presumably not life-supporting.

    I don't think your multiverse theory does this. Your theory just codifies your assumption that there are other possible universes that are not compatible with life, and tunes itself to our observations that our actual universe is one of the possible ones (actuality implies possibility).

    I guess I fail to see what is being explained. It is by definition that our universe is a possible universe because it is an actual universe. This is true, multiverse or no multiverse.

    Your multiverse theory has nothing to say about future experience. Nothing we ever see will increase or decrease our confidence in your hypothesis. Things would be different if we could peer into other universes, or find ourselves elsewhere, but this isn't possible. These fictitious universes are just a placebo, and I see no reason why something inherently unobservable should suddenly make us feel better about our state of knowledge.

    Let me explain this another way by analogy with mathematical constructions. Suppose I modify the traditional mathematical formulation of Newton's Laws by inventing new and more complex representations of physical objects and laws. For example, maybe space is filled with invisible springs, and matter consists of intersecting dodecahdra. However, my new formulation predicts nothing more than the original algebraic formultion. Would I be justified in claiming that I now have an explanation for Newton's Laws? I don't think so. My new theory is observationally totally equivalent to Newton's formulation. I would just have restated what I know about the world in a less economical form.

    Well, I think that generic multiverse theories are doing the same thing, only conceptually. You have tuned your model to generate what we know, but not to tell us anything new (at least nothing we can see).

    If such things count as explanations, then anything can be so explained just by cooking up some set of invisible things that jive with what we see, but which have no consequence for experience. For example, the many worlds interpretation of quantum mechanics shouldn't make us feel better about a car accident knowing that in an alternate universe the crash doesn't happen. I don't consider this explanatory because the other universes are undetectable in principle.

    One other point that Liu brought up. I don't think that anything I have said relies on science proper. I'm not demanding everything be scientific per se. I think my point should work equally well in philosophical or mathematical arguments. If we invent a symbolic representation of experience that claims to explain something, we can't later forget about the correspondence between those symbols and experience. There's no difference between non-existence and in-principle non-detectability.

  8. Ken, I don't think that response works, because it ignores the crucial role the anthropic principle plays in my explanation. After all, the existence of countless Shakespeare counterparts really doesn't do anything to explain why we have access to Shakespearean plays (which is presumably what really needs explaining here). There are presumably even more Shakespeare-less worlds that we might just as well have found ourselves in. But we could not have found ourselves in a lifeless world. So the multiverse + anthropic principle can provide explain the local existence of life, but not of other "particular improbable aspect[s] of our universe".

  9. My name is M.A.Carrano, and I'm a 27 year-old philosopher native to the Yale area New Haven, Connecticut. Hopefully, I could end this debate quickly with a peremptory remonstration of the multiverse hypothesis.

    My refutation is not going rely solely on empirical methods. This is so because the hypothesis itself is outside the empirical authority (which makes it a brand of faith) in terms of proving (though not in terms of disproving). Instead, I'm going to rely on a variant epistemology, logic, to demonstrate why the multiverse hypothesis is not only logically false, but also a brand of faith in the supernatural that is just as logically pernicious as Christian creationism.

    In order for the multiverse hypothesis to be true, relativity theory must be false. Or, in order for relativity theory to be true, the multiversal hypothesis must be false. This is so because the central axioms to either notion contradict with irreconcilable differences.

    According to relativity theory, time and space are natural forces. In juxtaposition with modern cosmology, time and space, like every other force in our universe, began with the Big Bang. Consequently, there was no time prior to the Big Bang, and there is no space outside of the boundary of the Big Bang - i.e. our expanding universe.

    This has some rather interesting implications with regard to the multiverse hypothesis. They are of a single kind: that the hypothesis is a patent violation of the laws of logic - namely the principle of non-contradiction; that, "In order for a multiverse to exist, a multiverse must not exist."

    Here's why.

    Since time and space are forces inextricable one another and from their parent universes, then there cannot be quantifiable separation of space and time between any individual universe within a given multiverse. More succinctly, because no space (and consequently no time) exists outside of our universe, every other universe in the multiverse must be coterminous with other universes, including our own, at exactly the same time. Therefore, the Multiverse hypothesis IS immediately verifiable. All it requires is that we set our telescopes to the edge of our universe and see if we can peer into those of our neighbors.

    Pretty neat, eh?

    So when we do this, what do we find? We see the boundaries of our universe expanding via the corresponding red and blue shifts of our stars into a pitch emptiness - not into other universes. Contemporaneously, we don't find other universes expanding into ours with the reciprocal tell-tale red and blue shifts of their own. As a result, the first form of the multiverse hypothesis is refuted.

    There is, however, one mode of a multiversal hypothesis remaining. However, it is equally if not more absurd than the first.

    Basically, since time and space are physical properties of their universe, then because neither time or space can exist outside of a universe (including our own), than every universe in a given multiverse must occupy the SAME SPACE at precisely the SAME TIME as every other. This is outright impossible. But let's entertain the notion further by simply declaring that they only do so virtually, and only in such a discreet fashion that they merely overlap one another without having to directly interfere with one another.

    Furthermore, let us entertain the idea even further by ignoring the irrationality of this argument by considering it worthy of testing regardless of its logical consistency. If this format for a multiverse were in fact legitimate, nevermind plausible, than the strength of gravity in our universe should reflect the the compounded gravitational influence of every other overlapping universe in the multiverse in the same range of space: our own. In otherwords, if there were an infinite ammount of universes overlapping our own, then the gravitational pull of our own should be infinite. Or, if there were trillions, billions, millions or simply thousands or hundreds of other universes overlapping our own, then the gravity of our own universe should reflect the combined gravitational influence of every other overlapping universe in the multiverse in our measurements of it. But since it doesn't, then this argument is also refuted. There are no overlapping universes parallel to our own...

    ...unless these other universes were non-physical universes completely alien from our own in terms of their elementals in every conceivable sphere of physical influence! Hence we stumble back into the problem of faith, only this one is faith in the supernatural - another logical absurdity.

    Conclusively, there are only three remaining options left for us to weigh between. Do we assume faith in a multiverse, despite its logical inconsistency? Do we concede that the Anthropic Principle is simply a brute fact - a "that's that" stroke of serendipity - that we accept on blind and unquestioning faith alone because science and logic have met their ends? Or do we continue to affirm the relevence of both and assert them with humility and intrepidition in the only remaining direction: in search for the Cosmic Architect behind our elegant universe?

    As for me personally? The choice is logical.

    Without prejudice,

    e-mail:, or visit me at:

  10. "because no space (and consequently no time) exists outside of our universe"

    This premise begs the question against multiverse theory. Obviously what the defender will say is that each universe has its own independent spacetime. It's not that two universes are located at some distance from each other in a larger "multiverse" -- that would just make them one big universe all along. Rather, the two universes have no spatiotemporal relations between each other at all. They are not close, nor are they distant. Neither exists "before" or "after" the other. They are simply wholly distinct.

    A "universe" is, I think, best defined as a maximal spatiotemporal region (i.e. for which all its parts are spatiotemporally related). So of course multiverse theory is "logically inconsistent" with the assumption that all things are spatiotemporally related. But that doesn't make it internally inconsistent. The multiverse theorist can simply reject this assumption of yours.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.