Beyond the Room

By Sean Carroll | July 14, 2008 10:21 am

I’m sure Ruben Bolling is making fun of people I disagree with, and not of me.

The underlying point is a good one, though, and one that is surprisingly hard for people thinking about cosmology to take to heart: without actually looking at it, there is no sensible a priori reasoning that can lead us to reliable knowledge about parts of the universe we haven’t observed. Einstein and Wheeler believed that the universe was closed and would someday recollapse, because a universe that was finite in time felt right to them. The universe doesn’t care what feels right, or what “we just can’t imagine”; so all possibilities should remain on the table.

On the other hand, that doesn’t mean we can’t draw reasonable a posteriori conclusions about the unobservable universe, if the stars align just right. That is, if we had a comprehensive theory of physics and cosmology that successfully passed a barrage of empirical tests here in the universe we do observe, and made unambiguous predictions for the universe that we don’t, it would not be crazy to take those predictions seriously.

We don’t have that theory yet, but we’re working on it. (Where “we” means an extremely tiny fraction of working scientists, who receive an extremely disproportionate amount of attention.)

CATEGORIZED UNDER: Humor, Philosophy, Science
  • Peter Woit

    “if we had a comprehensive theory of physics and cosmology that successfully passed a barrage of empirical tests here in the universe we do observe, and made unambiguous predictions for the universe that we don’t, it would not be crazy to take those predictions seriously.”

    While this is perfectly true, the fact of the matter is that the only such “comprehensive theory” being advertised (the string theory landscape) makes no testable predictions about anything, with no plausible reason to believe this is going to change. In the absence of such a “comprehensive theory” or any serious prospects of having one (and thus usable information about what kind of predictions it would make), discussion of the properties of an unobservable multiverse suffers from the problem of not being science in any conventional understanding of the term.

  • Sean

    You are reliable, I’ll give you that.

  • Bryan

    What a thoughtful comic.

    It seems to me that the fish should be saying, “I have good reason to believe that the predictions of my most reliable physical theories hold inside the room — and it doesn’t hurt to assume they hold outside, too.”

    That “good reason to believe” is that scientific reasoning is clearly ampliative, at least to some extent. The mature sciences reliably lead us from our (finite, local) experiences, to broader, more general, and sometimes universal conclusions about the world. But when can we say that it “doesn’t hurt” to do this?

    The really difficult thing is to determine 1) what kinds of things can be generalized, and 2) how far you can generalize them. As an example of the first: the fish’s conclusion is probably a bit liberal. But surely he’s justified in extending some of what he knows about (e.g.) freefall in the bowl, to freefall outside the room.

    As an example of the second: compare our confidence in the solar system predictions of classical-GR, to our confidence in its predictions outside or on a large scale. Sean, you’ve described disagreement about the latter in some of your lovely surveys on the accelerated expansion problem. It’s fascinating to me that some of these disagreements seem to turn on philosophical intuitions, about the scope of our ampliative reasoning.

    Best – Bryan.

  • John Merryman

    But tan the room has surfaces, like the bowl. So maybe the room is a bowl in a larger room! What is beyond that? Is it rooms all the way out?

  • Pieter Kok

    Isn’t there a window in the room, in the line-of-sight of the goldfish?

  • Mike

    Well, I think the cartoon makes physicists seem a bit more close-minded than they are. My impression is physicists are happy to admit the assumptions within which we work; only we tend to be quick to justify these because they are the only context in which we presently know how to work, and apparently they work quite well at describing the world around us.

    I don’t think the size of the universe is a good example of unimaginative deductions — anthropic landscape predictions are much more of this nature, though they get more than their due criticism for it.

  • B
  • greg

    I always enjoy trying to get people to explain to me what exists beyond the boundaries of the universe, or what the universe is like at it’s edges.

  • Count Iblis

    Well, the fish is more or less correct. You could say that the walls of the Room are domain walls separating different Rooms :)

  • Peter Lynds

    “Einstein and Wheeler believed that the universe was closed and would someday recollapse, because a universe that was finite in time felt right to them. The universe doesn’t care what feels right, or what “we just can’t imagine”…”

    It does care for ideas that result in real contradiction though Sean; it doesn’t follow them.

    Best wishes


  • Rumi

    The cartoon makes fun of cosmologists who claim to know what they say they do not know. Cosmologists admit that the unobservable part is unobservable. Otherwise it wouldn’t be unobservable. Then they invent countless academic doublespeak to claim that they know what they do not know.

    Here you claim that we can draw reasonable a posteriori conclusions about the unobserved from observation in the observed. For instance, the fish in the bowl can observe that humans exists outside the room. This is a verified observation. (Do you understand that humans exists outside the room does not mean humans exist in the entire universe???) A human comes into the room and feeds the fish. The philosopher fish can even reasonably a posteriori postulate that there are mean cats outside the room since there is that house cat staring at her with hungry eyes.

    Even going further the fish can draw reasonable conclusions that outside the room there is a moon and the moon wanes and waxes and there is even earth. He can even go further and listen to the humans in the room discussing cosmology and learn that there are planets and galaxies. All these pass a barrage of empirical tests. Fine.

    And as usual with this kinds of arguments presented by cosmologists such as this fish and you, at some point, a miracle must occur to let the cosmologists to make “unambigious predictions for the universe that we don’t observe.”

    Maybe you don’t understand. That there are humans outside the room, that there are galaxies outside the observed universe are not predictions about the whole universe. There exists unobservable universe. This is not even a problem about cosmology. This is a problem about language. Unobservable universe is unobserved. There is no ifs and buts about this. When they realize this cosmologists will become scientists. Now you guys are charlatans.

    Feel free to delete this comment. But it is very relevant that you qualified the predictions as “unambiguous” and not as provable. Unambiguous prediction is not a prediction because it cannot be tested. Remember unobserved will remain unobserved. So you dismiss Peter Woit as “you are reliable” but you are even more reliable in your promotion of cosmology as an untestable religion.

    I don’t think you even understand that what is unobservable is unobservable. And thanks for the cartoon. I remember reading a different version of this a few years ago when Stephen Hawking was big. The story goes like this. Hawking goes fishing. He falls asleep and his boat drifts into the ocean. When he senses a motion in the line he wakes up and pulls up the fish. It turns out that this fish is a cosmologist. They start talking shop. The fish says, the entire universe is made of water. This follows from the cosmological principle which says that the universe is homogenous and isotropic.

    Hawking laughs at the naivete of the fish who is now havinh trouble breathing. My dear colleague says Hawking, the universe is not made of water. Yes, the universe is homogeneous and isotropic, but it is made of galaxies. You just proved to yourself that there are media rarer that water, you cannot breath. As his last words the fish says, but in whatever direction we looked we saw water and nothing but water. Nothing but water. . . Nothing but water. . . Hawking says, don’t trust a posteriori unambiguous predictions cosmologists invent to furhter their careers, ha, ha, ha.

    The fish dies. Hawking says, mixing his metaphors, what a bird-brained fish. He observers water and thinks the entire universe is made of water.

    Do you think Hawking is right? Just remember this. When you die you don’t even know where you will be going. Organized religions invented a heaven and a hell for you to believe. They tell you that if you are good you will go to heaven if not to hell. You don’t believe them. But you are like the religious priests. Instead of heaven and hell, you create cosmoses and market them as scientific theories. Apply the same kind of skepticism that you apply to organized religion to cosmology as well. Humans are like the fish in a bowl. Or like a moth that lives but a spring. Only when you understand this you will become a scientist. Good luck to you. I am posting from the birthplace of Rumi from an internet cafe so don’t mind my language.

  • John Merryman

    A universe finite in time requires initiation, while an infinite one only requires inertia. That we observe a horizon line doesn’t prove the road comes to an end. If there were no boundaries to observe, would observation be possible?

  • Matt

    I think this comic would hit less of a sympathetic frequency in me if cosmologists made more of an effort to state their assumptions when talking to the layman. Every once in a great while, I hear that kind of qualification, say on the history channel’s “The Universe” (nice contributions, by the way, Sean), but they’re generally quickly followed with why those assumptions are perfectly reasonably. And they may be, I just want more of that.

  • Tyler

    i took that strip to be more of a comment on sophomore philosophy bull sessions than working cosmologists, but as they say, the guilty flee where no man pursueth.

    however, when working in such a field, i can see that – if one wants to make any meaningful progress in the scientific sense – one has to make certain grounding assumptions, as it were. and one could certainly make the *theoretical* assumption that the cosmological principle is correct (& all the other associated assumptions under discussion here) and proceed from there, without losing sight of the fact that these assumptions are simply that, axiomatic statements we must make at the outset, without which progress cannot be made. The only way this could be a serious problem is if the assumptions are elevated to the level of supposed factuality and given mental precedence over conflicting experimental data, or if the assumptions are *presented* as fact via the media to the public.

    This latter point is greatly complicated by the ongoing and apparently nearly total inability of the public at large to comprehend any amount of nuance when presented with scientific ideas. I’m sure this must be very frustrating to those doing the presenting. One can keep in mind, and even state, caveats such as “well we start from this assumption, and we have to keep in mind that this might not be correct and needs to be re-evaluated in light of any possible conflict with experimental data, etc” and either it will get edited out of the story, or the public will just think “blah blah blah don’t these poindexters ever say anything directly, do you know what you’re talking about or not?”

    Synchronistically, see

    Also: Ruben Bolling rules. This strip is good but pales in comparison to his sublime “God-Man” strips, collected here:

    Of which my favorite is “Is God-Man…Dead?” aka “God-Man vs. Nietzsche-Lad”

  • Lawrence B. Crowell

    Peter Woit: ” … (the string theory landscape) makes no testable predictions about anything, …

    Some RHIC data on quark-gluon plasmas indicates dualities with black hole interiors as expected with M-branes. So the this statement might have a few little cracks in it. The big problem with testing string theory is the same as with LQG, the scales where the theories operate are utterly extreme. Both theories might just be different perspectives on these problems, and both might have some elements of physical reality.

    Lawrence B. Crowell

  • Peter Woit


    The problem with testing string theory is not that you can’t test its predictions since they are at the Planck scale, it’s that there are no predictions at any energy scale. This includes making no predictions about what the ground state and low-energy excitations about it look like.

    The question of whether string theory provides a useful model that can be compared to the RHIC data has nothing at all to do with whether it provides a predictive TOE.

  • Lawrence B. Crowell

    The fish’s question, “What could possibly be beyond the room?” means that there is some sense of the beyond. The question in a sense calls for at least some imagination of the beyond. Of course many of these a posteriori conclusion are likely to be wrong” It’s turtles all the way down.” But even if they are wrong they still have value, for if or when we do actually observe the beyond we have something to compare with what we see.

    Lawrence B. Crowell

  • Lawrence B. Crowell


    One of the terms that gets tossed about in physics that I have become unhappy with is TOE. Maybe it’s better to use TOS, Theory of Something. The RHIC data does suggest a possible signpost on string theory that illuminates some neighborhood in the theory. This does not mean string theory is in anyway confirmed (science never really confirms theories), but that nature might share some features common to strings.

    String theory fails to “constrain” very well. As a result we have a vast multiplicity of “theories” on how E_8xE_8 string decompose on a vast number of vacua and a seemingly endless nesting of more complicated algebraic geometric constructions. I will say that for various reasons I think there are stringy aspects to this domain of physics. This does not mean I think string theory is the final answer to things. It appears that it can’t be, for something else appears required, or else the whole theory is simply wrong.

    I am of the opinion that some basic principle is required. I suspect that some of the difficulties within theoretical physics is that some very basic idea is lacking in what we are doing. Some sort of new physical principle I suspect is required to make better sense of what we are working on.

    Lawrence B. Crowell

  • Tyler

    Lawrence wrote:

    “Some sort of new physical principle I suspect is required to make better sense of what we are working on.”

    Interesting how much more often I see some variety of this sentiment expressed these days compared to, say, 5 years ago. Not counting crackpots.

  • Chris W.

    The universe doesn’t care what feels right, or what “we just can’t imagine”; so all possibilities should remain on the table.

    The great thing about assuming homogeneity and isotropy is that the assumption has consequences that can be tested. Thus, it could turn out to be wrong! What we have observed, and even what we imagine we can observe in principle, have changed over time. Every time this happens, apparently simplistic assumptions like homogeneity and isotropy come under threat. What is amazing is how often they hold up.

    It therefore sometimes pays to make unjustified but testable assumptions that strike some, perhaps may, people as implausible, if somebody is willing to test them. Of course, if the culture of science evolves away from this—eg, for reasons of cost or widespread anxiety about career prospects—then its development will be profoundly impaired in the long run.

    It doesn’t matter whether or not Einstein and Wheeler thought the assumption of a spatially and temporally finite universe “felt right”, if the idea constrained observational outcomes. Limitations on expected observations is what we need; because we learn something when those expectations are violated. Although it isn’t clear what its promoters intend, the idea of the multiverse smacks of leaving “all possibilities…on the table” in a single theory, thus ensuring that such a theory can never be contradicted by observation. If one supposes that the most open-minded and ultimately scientific attitude is to consider all possibilities, and one would also like to be right, then it is tempting to build all possibilities into a single theoretical framework. Unfortunately, in pursuing this goal science devolves into metaphysics—a metaphysics full of fascinating and imaginative mathematical (and other) ideas and constructions perhaps, but still metaphysics.

    Of course one can still say: “Perhaps that is just the way the universe [multiverse] is.” As a purely logical matter there is no good response to this. As a matter of pure logic, the universe presumably can be anything its wants to be, including something we have no hope of grasping scientifically. For some strange reason, the universe has cut us a break:

    The eternal mystery of the world is its comprehensibility.

  • Chris W.

    PS: Karl Popper once offered an observation and an admonition, neither of which are often mentioned. The observation was that science, viewed in his terms as a process of conjecture and refutation, would lose its way without some measure of success—recurring production of conjectures that stand up to careful tests over some significant period of time.

    Popper’s admonition was that science should not attempt to explain its own success; that way lies a kind of damnation (my word, not his). I suspect that, in part, he had in mind Kant’s attempt to account for what he (Kant) imagined was the indubitable and final success of Newtonian mechanics, some decades before it became fully apparent that its success was neither indubitable or final.

  • Boltzmann’s Reptilian Brain

    Sean Said:
    “You are reliable, I’ll give you that.”

    Dear Sean: there is a vast gulf fixed between “reliable” and “endlessly repetitive, with no substantial argument”.

    I also strongly disagree with your closing line about
    “extremely disproportionate amount of attention”. What’s disproportionate about it? It’s good when interesting things receive more attention than boring things, and this is a canonical example. Examples of things that really *do* receive more attention than they are worth are easy to come by: nano-anything, medical “research”, etc etc etc.

  • Peter Woit


    Whoever you are, I don’t recall you ever making a substantive argument of any kind here.

    As for my comment and Sean’s response, I’ll let others judge which was more substantive.

  • Greg Egan

    Rumi wrote:

    I am posting from the birthplace of Rumi from an internet cafe so don’t mind my language.

    Ingilisi-ye-shoma kamil ast. Shoma dar Balkh hastid?

    I think questions like this can be divided into cases of various degrees of confidence. I’d bet every cent I own that classical general relativity continues to hold just as well a few metres inside the event horizon of a large black hole as it does outside the hole (and for a large enough hole someone could even take me up on this bet; we could jump in side by side, and the loser could hand over a small ingot of gold to the winner before dying).

    And I agree when Sean says:

    it would not be crazy to take those predictions seriously

    Certainly not crazy. Still, when it comes to making claims about a literally infinite unobservable region of the universe, based on our observations of a finite region, I’d certainly be much less confident than I would be about the interior of a black hole.

  • John Merryman

    Tyler on Jul 14th, 2008 at 5:17 pm
    Lawrence wrote:

    “Some sort of new physical principle I suspect is required to make better sense of what we are working on.”

    Interesting how much more often I see some variety of this sentiment expressed these days compared to, say, 5 years ago. Not counting crackpots.

    The reason revolutions do tend to originate on the margins and in the cracks is because growth is bottom up, be it biology, geology, economics, politics, etc. So long as earlier process can contain and define later process, the structure continues to grow, but when top down pressure and bottom up growth create a disconnect, start looking for a new model.

  • daisyrose

    What makes any work of art is a sense that there is something beyond – beyond what is observable – infinity – Parts of the forest you did not know existed before – its a mystery – you cant see it but you know there is something more out there.

  • Sean

    Okay, now I’m being made fun of. In a good way.

  • Neil B.

    I thought one of the problems was whether laws varied from place to place (like the rules do on Earth!). Otherwise, it’s merely a matter (heh) of how far “the same stuff” – including dark matter/energy – extends. But how can anyone get a handle on “basis for laws” without invoking speculation about First Causes, or just saying we’re in one of a multitude of universes – with no proof? Some talk of strings being a basis for variation in laws either from place to place in a big contiguous space, or in separate spaces (“universes.”) But as I’ve asked before: we can predict what a real violin etc. string will do because it’s made of atoms, and we invoke the fundamental properties of those constituents. But what explains what strings (or particles without parts if you don’t believe in strings, like muons) act like? If they aren’t “made of something” and we must presumably take their behavior as given fundamentally, but that’s just a command economy type thing. It has little explanatory value since there isn’t an underlying, even more fundamental interaction inside to model it with, to build up from.

  • Anisotropie


    I was wondering what do you think about CMB-based tests (like Integrated Sachs-Wolfe, Cold Spot) for string theory and braneworld models.

    With future gains in the uncertainties and error bars in the correlation between CMB and LSS (as Mark said here we can use evolving potentials to make predictions on f(R) models, and maybe (imho) on braneworld models too.

    Thank you very much

  • John Merryman


    Maybe it’s a little of the nature vs. nurture dichotomy. That it isn’t just the fundamental nature as cause of everything arising from it, but the element of feedback is integral to what the nature of the fundamentals are in the first place? Much more confusing of course, with more of an eternal cycle, than first cause.

    Interesting article on E. O. Wilson’s latest doings that brought the observation to mind;

  • lauren kennedy

    It seems to me, in most cases, scientists or anyone for that matter, start with an unproven idea and then set out to prove it. Often, it may be decades latter that it can actually be proven. Our technology must first extent our perception abilities.

    If we could communicate with some of the animals, they would have told us long ago that an ultra violet spectrum or above the magnetic energy around the earth because they could perceive them. The point is that just because we don’t perceive something, doesn’t mean it’s not there.

    Then of course, there is the huge problem of perception itself. Our interpretation of what we do glimpse in the Universe may be very skewed. We make interpretations based on what we experience, believe or imagine. Plus, since the observer cannot observe without producing some effect on what he has observed, how can you know for certain what is there when it isn’t being observed?

    Since everything is made of energy, even our thoughts and imaginings, and the smaller the sub atomic particles or proposed particles that compose it, the more unpredictable it becomes, how can you know what affects the particles of energy that our thoughts contain and the particles of other energies zipping around have on each other? I suspect that there may be something even more basic and less tangible than energy. That would take quite an experiment to prove.

    Interestingly, some of the imaginings of many of the ancient mystics or elite of some ancient spiritual theories, have much in common with many of the scientific theories being proposed or proven today.

    It seems to me that theories of gravity still have some gaps and we thought we had that all sewn up because Newton had proven that it worked and that it could be used to predict how things moved. Then Einstein revised that, and everyone thought it was a done deal again. But then quantum physic came along and some of the newer observation don’t always seem to fit as well as one would like
    Why is it so weak compared to the other forces. No one knows.

    The point is that is it presumptuous to every assume that we have it figured out. Not that we shouldn’t try. But imagination must come before theories and theories before proof and proof before anyone considers it a fact. But facts also can be presumptuous. Until we have the senses to see everything and the wisdom and awareness to fully and accurately understand what we are seeing, any proven fact is just a highly effective theory.

    So quit the back biting. All the disciplines have something to ofter. And people generally believe what they want to believe, “facts” or not. However, I do think anything should be presented accurately, pro and cons, results – one that prove or disprove.

    Especially for lay people like me who are fascinated, curious and intrigued by new ideas, information and discoveries but lack the technical education to understand equations.

    You may dismiss these observations from a mere lay person if you like. I am still eager for all the new information that I can get as the result of your pioneering research and discoveries. It is exciting, intriguing and fascinating. nd the Universe is amazing.



Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

About Sean Carroll

Sean Carroll is a Senior Research Associate in the Department of Physics at the California Institute of Technology. His research interests include theoretical aspects of cosmology, field theory, and gravitation. His most recent book is The Particle at the End of the Universe, about the Large Hadron Collider and the search for the Higgs boson. Here are some of his favorite blog posts, home page, and email: carroll [at] .


See More

Collapse bottom bar