Endless Universe

There’s a new popular book about cosmology now out on bookstore shelves, Endless Universe by Paul Steinhardt and Neil Turok. The authors are the inventors of a competing model to inflationary cosmology, variously called “ekpyrotic” or “cyclic” cosmology. They describe coming up with the same idea simultaneously during a lecture at Cambridge on M-theory by Burt Ovrut back in 1999. Ovrut was describing his work on the Horava-Witten scenario, which involves two parallel branes (we live on one of them), and during the lecture both Steinhardt and Turok started wondering about whether one could explain the big bang as a collision of branes. They went up to discuss this with him after the talk, and continued the discussion on a train ride to London that evening to see a performance of the play Copenhagen. This train ride was a central part of a 2002 BBC TV program Parallel Universes and a recent play Strings by Carole Bugge that I saw performed here in New York late last year.

The book is very much an advertisement for cyclic cosmology, and devotes a lot of space to doing something which is rarely done, explaining the problems with inflationary cosmology. Steinhardt has worked extensively on coming up with viable inflationary models, and a large part of the book explains the story of this research. Perhaps the most interesting aspect of the book is the story that Steinhardt and Turok each tell about their careers and how they ended up working together on this alternative to inflation. The problems with inflation are described in the context of promoting their cyclic model of branes colliding, moving apart and then back together in a repeating pattern. They heavily sell the idea that a cyclic model with no beginning of time is conceptually much preferable to a standard inflationary model in which the universe emerges at a given time. Reading this made clear to me why, at the recent String Cosmology meeting here in New York, Turok was so persistently questioning one speaker about whether everything he was doing didn’t just depend on an unmotivated choice of initial conditions.

Steinhardt and Turok do a pretty good job of demolishing the inflationary multiverse and the associated Anthropic Landscape philosophy that has become so popular in recent years. They correctly describe the main problem with the inflationary multiverse as the lack of any way to test the idea, even in principle, making it not really a scientific idea at all. As for the testability of their own theory, they devote an entire chapter to the question of contrasting its predictions for the CMB with those of inflation. They claim that both cyclic and inflationary cosmology make the same predictions for the WMAP results (implicitly criticizing the commonly made argument that WMAP provided strong support for inflation). The one possible test that they point to that could distinguish the inflation and cyclic scenarios is the expected more sensitive measurement in coming years of a possible B-mode polarization signal due to gravity waves in the CMB. They claim that inflation predicts a significant amount of B-mode polarization, whereas the cyclic model doesn’t. Unfortunately, from what I can tell by looking at the recent literature on “string cosmology” (e.g. here), various inflationary scenarios can give a wide range of amounts of such polarization, with stringy models like “brane inflation” and “modular inflation” leading to essentially none, just like in the cyclic case. So, I guess the cyclic model is in principle falsifiable, if next generation CMB experiments turn up measurable B-mode polarization. But if this doesn’t happen, I don’t see how one is ever, even in principle, going to distinguish experimentally between the cyclic and inflationary scenarios, which will make this whole area of research highly problematic.

On the whole the book seems to me to be too much of an advertisement for a very speculative idea, and I don’t think the public needs more of this in this kind of format. I didn’t notice anything in the book about what the case against cyclic cosmology might be, so anyone who wants to find out the other side would have to go on a search of the scientific literature, something most members of the public might not be able to do. Most strikingly, since the cyclic model is based on brane ideas motivated by string theory, the book contains endless hype about string theory, without so much as a word about its problems. One would have to read extremely carefully to realize that there is not a shred of experimental evidence for string theory. As for recent public debates about the problems of string theory, the authors just pretend they don’t exist. They give a long list of recent popular books in this area, such as those of Susskind and Vilenkin promoting the Landscape, but somehow neglect to include the two such books that have taken a critical point of view on string theory.

Associated with publication of the book, it looks like there will be various stories in the media promoting the cyclic vs. inflationary debate, trying to make it into a modern version of the old steady-state vs. Big Bang controversy. On the Edge web-site there’s a recent piece by Turok, which gives a good explanation of his current research and point of view. From NPR, there’s a very recent radio show about the cyclic model, entitled Forget the Big Bang Theory, where “renegade physicist” Turok’s model is described as “fighting words in the halls of science.”

This entry was posted in Book Reviews. Bookmark the permalink.

29 Responses to Endless Universe

  1. anon. says:

    They claim that inflation predicts a significant amount of B-mode polarization

    Many people would disagree with that. The well-known Lyth bound suggests that observable B-mode polarization implies Planckian field values. This is extremely difficult to obtain in string theory; see work by Baumann and McAllister, for instance. It’s also not very plausible in effective field theory, where one expects potentials to get Planck-suppressed corrections and flatness at trans-Planckian scales would be difficult to explain. There have been attempts to circumvent this with, e.g., inflation driven by large numbers of axions, but nothing very compelling. An observation of large tensor signal would cause a real theoretical crisis, I think.

  2. Kris Krogh says:


    Glad to see you shine a little light on cosmology, another near-monopoly situation with many parallels to string theory. Who says the Big Bang is “the only game in town?”

  3. Coin says:

    Who says the Big Bang is “the only game in town?”

    Ah, be careful, that’s the kind of statement that you’ll tend to see quote-mined on Talk.Origins later 🙂

    My understanding of the “cyclic universe” model described here is that it does not supplant the traditional big bang model (at least the big bang model as the layman would understand it), but rather extends it. Is this not correct?

  4. Kris Krogh says:

    Hi Coin,

    Haven’t read the book, but the universe had a beginning in BB. Somehow “Endless Universe” doesn’t sound like that to me.

  5. M says:

    Inflation predicted that primordial inhomogeneities are adiabatic, Gaussian and with a quasi scale-invariant spectrum. All of this was later confirmed by WMAP and other observations.

    Bounce cosmologies have problems in postdicting these features. First, cosmological singularities in string theory are not understood well enough to know if a bounce really occurrs. Second, perturbations generated “before” the bounce do not need to be scale-invariant, in gross disagreement with observations (maybe some miracle happens during the bounce: we do not understand this critical phase). This can be fixed in specific models: by assuming ad-hoc potentials, by introducing extra fields.

    This is why many cosmologists do not consider ekpyrosis as a serious competitor to inflation. And somebody jokes that ekpyrosis sounds like a beautiful name for a new illness…

  6. Alex Nichols says:

    ‘Ekpyrosis’ is actually an old term derived from the Stoic philosophers, implying that the universe was created “out of fire”.
    Certainly, the observational evidence implies that the early universe was very hot.

    S & T also revived the term ‘Quintessence’ in their cyclic model, originally the Aristotelean “5th element”, through which the stars and planets were believed to move.
    They use it as an explanation for ‘dark energy’, suggesting that it’s a slowing rolling force (currently increasing), which explains recent obervations on cosmic expansion.

    Cyclic Cosomologies were quite popular in the early 20’s, based on a closed universe with a negative cosmological constant.
    For example, the Friedmann, Walker, Roberton, Le Maitre models.

    In the 1930’s, Richard Tolman showed that this would lead to an entropy build-up, so that each cycle would become larger than the previous one, eventually leading to a situation where gravitational attraction would be insufficient to cause re-contraction. This severely reduced the model’s explanatory power and it went out of favour.

    Then in the 1960’s Penrose and Hawking showed that gravitational singularities would form in any big-bang scenario based on General Relativity. This would erase the information from previous cycles. They didn’t accept that there was any evidence that the “big-bang” was preceded by a “big crunch”. Penrose in particular, has argued that there is no reason why a ‘crunch’ would lead to a ‘bounce’.

    As current thinking is that the cosmological constant has a small positive value, gravitational cyclic models based on General Relativity and the Friedmann equations are not in favour.
    A quantum or string based description of the events approaching a big bang singularity would be required to explain what actually caused such an event.
    This is what Steinhardt and Turok are attempting to provide with the idea of a brane collision.

    So far, it hasn’t been disproved by the WMAP results and future experiments such as the Planck satellite should be able to provide tests of the polarisation of the CMB and Spectral tilt.

    A more radical view of the situation would be that it’s not meaningful to try and explain ‘the universe’ as a physical system and that valid physics can only be performed locally.

    In such a view there is an extremal indentity between the singularity and cosmic horizon. As one approaches either, quantum and or string effects are essential in explaining what happens, but it’s only possible to infer these theoretically, rather than directly measure them.

    S & T certainly do list a series of observational consequences of their model that can be tested.

  7. island says:

    Then in the 1960’s Penrose and Hawking showed that gravitational singularities would form in any big-bang scenario based on General Relativity. This would erase the information from previous cycles.

    Interesting, if you project relativity backwards to the point where inflation is alleged to end, then that is also where you conveniently find “thermalization”, reheating or, “Ekpyrosis”.

    The singularity arrises only if this isn’t indicitive of a big bang in a universe that has pre-existing volume… which would *not* erase information from previous cycles.

  8. Peter Woit says:


    At least in the book, their claim is that their model has the same observational consequences as inflation, except for the CMB polarization.

  9. Navneeth says:

    Associated with publication of the book, it looks like there will be various stories in the media promoting the cyclic vs. inflationary debate, trying to make it into a modern version of the old steady-state vs. Big Bang controversy.

    It has made it to Slashdot. 🙂

  10. matteoeo says:

    I’m an italian physics undergraduate student. Although I took several classes in GR and cosmology, I’m no expert. But it doesn’t take an expert to understand that the state of cosmology is pretty messy. I think that there is some very smart speculation going on, but on the whole it seems to me that cosmology is too young a science to really tell something about the universe. What mostly strikes me is that cosmology uses as ingredients all kinds of speculative (sometimes unscientific) hyphotesis and still unknown objects: higgsboson, SUSY breaking, superpartners, strings, branes, dark matter, dark energy, CC, quintessence, extended quintessence, GUTs an GUT simmetry breaking, and so on. As a result you can adjust inflation so to have almost any CMB, but this is not really scientific. If every ingredient had a 50% possibility to come out true, which is the probability that cosmology makes sense at all?
    I’ll tell you why I feel so uncomfortable with cosmology. A few years ago italian cosmologists and theoreticians Matarrese, Riotto, Notari and american Kolb showed (here) that today’s accelerated expansion might be explained as the effect of inhomogeneities in the matter distribuition (super-Hubble wavelenght fluctuation modes), which combined with standard GR act as a cosmological costant in the Friedmann equations (they assume inflation). So standard GR and a better comprehension of complex behaviour of matter coming out from the fact that the universe is not homogeneous at all might rule out ad hoc-CC and DE. But what is most interesting is to read the thread of arguments that these pepers raised, with all kinds of objections, mostly of a very general flavor: no go theorems for irrotational matter to cause negative pressure (doesn’t apply to their case though), a discussion on whether or not these effects might suffice to give negative pressure and the need for a non-perturbative approach, whether or not one can use Friedmann equations and modify them while assuming an inhomogeneus universe and so on.
    I’m not publicizing their idea, which I cannot evaluate; but I strongly believe that this whole thing shows that the very foundations of Cosmology (Cosmological Principle above all) are not understood and questionable, that all sorts of very general problems are not at all estabilished, and thus work in the field is easy to criticize (cosmologists dismantling others’proposals). I also think that before assuming all kinds of hyphothesis a much better understanding of GR dynamics, complex systems and collecive phenomena should be gained.

  11. Alex Nichols says:

    re Penrose/Hawking:-

    During the 1990’s Penrose suggested that the boundary condition of a collapsing universe could be distinguished by a large Weyl tensor.
    Hawking agreed but said it was determined by the no boundary proposal.
    Turok later worked with Hawking on an inflationary model based on this, but has moved on since then.

  12. Malo Juevo says:

    The paper by Kolb et al. was known to be obviously wrong from the moment it appeared. Working out precisely what the error was turned out to be not so easy, but perturbations around Robertson-Walker spacetime have been well understood long enough that something like this would not have been missed. The calculations in the paper are quite intricate, making it very difficult to keep track of what the range of validity of perturbation theory is at each step. However, if one looks at it very carefully, the perturbative technique is not valid. In some sense, this might be expected, since the whole point of the paper is to claim that a certain variety of perturbations are singular, yet the paper is done entirely in non-singular perturbation theory.

  13. Peter Woit says:

    If you have informed comments about Steinhardt-Turok, please post them. But this is not a general Cosmology discussion board, and I can’t and won’t moderate such a thing, so if it’s not about Steinhardt-Turok or their book, don’t post it here.

  14. Ari Heikkinen says:


    It appears to me that “Endless Universe” is more like speculation on a new theory than it’s cosmology, so could you suggest a good university level book (preferably self contained) on modern cosmology which would include all the relevant observations of the universe and their related math, but in a manner that it would explain what’s seen, without trying to speculate on any specific theory?

  15. Cynthia says:

    In the “Edge” transcript, Turok uses the term “_mdash” not once but three times: “realistic&mdashspoils”, “bring&mdashlike”, and “argument&mdashhand”. Although the meaning of the term “m&dash” is pretty obvious (to me, that is), does anyone know if it’s really proper to be using this term in essay format?

  16. Chris W. says:


    The occurrences of &mdash in the text are due to sloppy HTML editing on somebody’s part; I doubt Turok had anything directly to do with it. — (—) is an HTML 4.0 character entity.

    [Example: Web developers do make mistakes—sometimes.]

    See this reference.

  17. clarity says:


    I just want to make a few distinctions which are, I think, a little muddy in your posting. First off: inflation, brane inflation and the multiverse/anthropic landscape are all distinct. The simplest inflationary scenario is very well supported by the data and is certainly testable (it’s already been tested!). Particular string theory embeddings of inflation (such as brane inflation) may or may not be falsifiable, though there isn’t currently a way to rule out inflation from string theory (however, a positive signal for tensor modes would give stringy inflation a hard time). Eternal inflation and the multiverse are both controversial, even among proponents of inflation and are certainly not necessary parts of the puzzle. Inflation doesn’t necessarily have anything to do with anthropic arguments and many cosmologists dislike anthropism as much as you do.

  18. Peter Woit says:


    Sorry if I wasn’t clear, I am aware of the distinctions you mention, and that simple inflationary scenarios do have some real experimental support (as well as problems that Steinhardt and Turok do a good job of describing). I didn’t mean hear to be criticizing this kind of thing, I think it’s serious science (brane inflation and eternal inflation/multiverse are a different story).

  19. clarity says:


    Thanks, I just wanted to be sure that you weren’t trying to lump all those things in together. The problems with inflation are largely conceptual and, to my mind, many of these problems have deep quantum gravity issues at their core. Personally (you are likely to disagree with this) I think that it will be hard to completely resolve all the problems of simple inflationary scenarios without coming up with some kind of UV complete, quantum gravity version of the story. So in this sense I don’t think that brane inflation (and similar scenarios) are nearly as poorly motivated as you think they are. I’m not trying to argue that any of these string theory inflation models actually solves any of the outstanding problems of inflation, just that trying to embed inflation into a quantum theory of gravity is probably a step in the right direction. (I should probably mention that I don’t think that the ekpyrotic/cyclic scenarios offer much, if any, resolution to these problems, but that’s another story entirely…)

  20. Peter Woit says:


    My own general view is that to understand inflation you will need to understand the beyond Standard Model physics that is driving it. Quite possibly this requires a good theory of quantum gravity. Historically I guess the hope was that maybe inflation came from some kind of Higgs sector, but I gather that doesn’t work.

    The problem with all string theory inflation scenarios that I’ve seen seems to me to be exactly the same as the problem string theory has with getting particle physics: compactification. You have to invoke complicated compactifications in order to avoid conflict with what is known, and this ruins predictivity. I’ve looked a bit recently at brane inflation papers (including a recent Klebanov et. al. one this evening), and the constructions just look complicated, ugly, and with no hope of leading to real predictions.

  21. clarity says:


    “My own general view is that to understand inflation you will need to understand the beyond Standard Model physics that is driving it. Quite possibly this requires a good theory of quantum gravity.”

    We agree on this for sure. The thing that makes the quest to embed inflation into beyond-the-standard-model physics interesting is the fact that inflation is not at all a generic property of sensible theories of high energy physics. Typically the requirement of sufficiently many e-foldings puts stringent (often prohibitive) constraints on the underlying theory. This is one of the main points of the recent Baumann et al. papers – it’s very difficult to tune the potential to be flat enough (even though a naive parameter counting suggests it should be simple). The fact that it’s so difficult to come up with concrete, working models of inflation in string theory suggests, to me, that the models which do work will be at least somewhat predictive. (I’m still working through the paper you mention so I won’t comment on how falsifiable this scenario is, though I think that a large tensor signal would rule it out for sure.)

    Of course, as you mention, there are aesthetic issues and I won’t claim that these constructions aren’t complicated. It’s not clear to me how complicated is “too complicated” when dealing with these kinds of theories, there’s certainly some personal prejudice involved in making that call. Given that it’s far from obvious that one can construct concrete, explicit, successful inflationary models in string theory, I think that any example (even a complicated one) is interesting.

  22. Peter Woit says:


    I think even the authors of the paper I mentioned acknowledge how problematic the whole thing is, since they end their paper with:

    “Finally there is a pressing need for a more natural model of string inflation than the one we have presented here”.

    which is equivalent to saying that they believe the model they are working with is highly unnatural and clearly they have little faith that it reflects the real world.

    Personally, looking at the calculations done in this and similar papers, the whole thing seems to me completely absurd. You’re invoking a huge amount of complicated structure, and trying to explain just a few observable numbers. This is hopeless. It’s not an aesthetic question, it’s a question of whether the number of assumptions that go into your calculation is larger than the number of observable predictions you can derive from it, and thus whether your framework is predictive.

    Falsifiability is not enough to make a legitimate testable scientific theory. My theory that everything that happens in the universe happens because it is controlled by a malevolent being with 11 toes is falsifiable, because an omnipotent malevolent being with 10 toes may put in a public appearance at any moment. The fact that a complicated cosmological model predicts no measurable effects of primordial gravitational waves is not enough to make it a testable model in the usual sense. You have to come up with measurable effects that are characteristic of the model, and distinguish it from others. I don’t see anything like that even conceivable with the string inflation models people are working on.

  23. wow says:

    Several experts now think cosmic strings from brane inflation (admittedly a very model dependent phenomena) may be distinguishable from the traditional abelian Higgs model networks due to different properties of intercommutation as well as (p,q) flavor structure; this is not obviously true or false, but is in any case a potential distinguishing signature that experts are sorting out.

    A similar thing could be said about non-Gaussian signatures that
    arise from inflation with D-branes governed by the DBI action, in
    the regime where the kinetic structure is important (not slow roll).
    Experts on non-Gaussianity (including the WMAP team) use this
    model as a reference point to bound non-Gaussianities.

    So I think your claim that characteristic measurable effects
    aren’t “even conceivable” is not held by most of the people who
    actually try to build inflation models, or test them experimentally.

  24. Peter Woit says:


    Your first example starts out “assume we observe a network of cosmic strings…”, something which most people would describe as an extremely unlikely event. From there you go on to note that this is a very model-dependent phenomenon and it’s not clear if you can even distinguish it from non-string theory cosmic strings. Sorry, this just doesn’t sound to me like a plausible test of brane inflation models: you’re starting assuming something that isn’t going to happen, then acknowledging that even assuming this, you can’t necessarily make any predictions.

    Your second example again assumes that we observe something that conflicts with simple inflationary models. Fine, but is this really a distinctive signal that one is seeing brane inflation? I can’t help but suspect that if such an effect is observed, people will have no trouble coming up with many very different models that would explain it.

  25. wow says:

    By definition, a “distinctive” signature has to be something that
    is not common to almost all models. So such things will be, at some level, unlikely. I suspect most working theoretical cosmologists would assign significantly greater probability to observing a cosmic string network, or non-Gaussianity, than they would to theoretical consistency (let alone experimental verification) of the Ekpyrotic/Cyclic scenario. I agree that most would also suspect absence of all these things; but that is just saying that vanilla inflation will look good, and we won’t know the correct model at the level of a very specific potential or (in higher dimensional scenarios) compactification manifold.

    My point was that signatures with roughly this likelihood constitute a major portion of what phenomenologists work on/hope for, and experimentalists try to measure, since by virtue of their somewhat rare nature they are very instructive when found. And it is quite conceivable to a majority of active workers in theoretical cosmology that such signatures are forthcoming sooner or later, to the extent that major current experiments use models motivated in this way as benchmarks. For me this puts your claim that there are “no conceivable” singatures of such models, in a more realistic context.

  26. clarity says:


    You’re way off about wow’s comment about large nongaussianity. Many theorists have worked very hard to come up with inflationary models which have large nongaussianity and there are, currently, very few examples. Among those few examples most of them could be observationally distinguished by the shape and running of the nongaussianities which contain a wealth of information about the underlying theory. So a future observation of large nongaussianity (which is entirely possible) which is consistent with the results of DBI models would certainly be strong evidence for brane inflation. This is the position of most people who work on cosmological nongaussianity. It’s not true that many other models could explain the same observation.

  27. Peter Woit says:


    I’m making no claim to expertise in this area, so will take your word that there are few ways to get non-gaussianity. Can’t help though being suspicious that if it’s ever observed, people will find more….

  28. clarity says:


    Of course there might be more ways to get big nongaussianity that nobody has thought of yet. I would be surprised if there are lots more ways, however, because there has been a lot of activity in trying to construct models with f_NL >> 1 and the list of models considered seems pretty exhaustive to me. It’s clear that large nongaussianity requires something pretty novel to occur either during inflation or else shortly afterwards.

    Also, it’s worth pointing out that the bispectrum constains a lot of information and is rather model-dependent so if, as you say, many more models with big nongaussianity are discovered it still seems likely that they’ll be (at least in principle) distinguishable from DBI.

  29. Pingback: Not Even Wrong » Blog Archive » Less Stuff Than Usual

Comments are closed.