A Cosmic Controversy

A couple months ago Scientific American published an article by Ijjas, Steinhardt and Loeb (also available here), which I discussed a bit here. One aspect of the article was its strong challenge to multiverse mania, calling it the “multimess” and accusing multiverse explanations of being untestable and unscientific.

Yesterday Scientific American published, under the title A Cosmic Controversy, a rebuttal signed by 33 physicists, together with a response from the authors, who have also set up a webpage giving further details of their response. Undark has an article covering this: A Debate Over Cosmic Inflation (and Editing at Scientific American) Gets Heated.

As Ijjas, Steinhardt and Loeb point out on their webpage, the story of this letter is rather unusual. It was written by David Kaiser and three physicists well-known for their outspoken promotion of the multiverse (Guth, Linde and Nomura). Evidently these authors decided they needed reputational support on their side, and sought backing from other prominent names in the field (I’m curious to know who may have refused to sign if asked…). Their letter starts out with a claim to represent the “dominant paradigm in cosmology” and notes the large number of papers and researchers involved in studying inflation.

If you read carefully both sides (IS&L and GKL&N) of this, I think you’ll find that they are to a large degree speaking past each other, with a major problem that of imprecision in what one means by “inflation”. To the extent that there is a specific identifiable scientific disagreement, it’s about whether Planck data confirms predictions of the “simplest inflationary models.” IS&L write:

The Planck satellite results—a combination of an unexpectedly small (few percent) deviation from perfect scale invariance in the pattern of hot and colds spots in the CMB and the failure to detect cosmic gravitational waves—are stunning. For the first time in more than 30 years, the simplest inflationary models, including those described in standard textbooks, are strongly disfavored by observations.

whereas GKL&N respond:

there is a very simple class of inflationary models (technically, “single-field slow-roll” models) that all give very similar predictions for most observable quantities—predictions that were clearly enunciated decades ago. These “standard” inflationary models form a well-defined class that has been studied extensively. (IS&L have expressed strong opinions about what they consider to be the simplest models within this class, but simplicity is subjective, and we see no reason to restrict attention to such a narrow subclass.) Some of the standard inflationary models have now been ruled out by precise empirical data, and this is part of the desirable process of using observation to thin out the set of viable models. But many models in this class continue to be very successful empirically.

I take this as admission that IS&L are right that some predictions of widely advertised inflationary models have been falsified. Of course, if these had worked they would have been heavily promoted as “smoking gun” proof of inflation, as was demonstrated by the BICEP2 B-mode fiasco. After BICEP2 announced (incorrectly) evidence for B-modes, Linde claimed this was a “smoking gun” for inflation (see here) and the New York Times had a front page story about the “smoking gun” confirmation of inflation vindicating the ideas of Guth and Linde. A couple months later, before the BICEP2 result was shown to be mistaken, Guth, Linde and Starobinsky were awarded the $1 million Kavli Prize in Astrophysics.

GKL&N don’t mention the sorry story of the BICEP2 B-modes, what they have to say about this is

the levels of B-modes, which are a measure of gravitational radiation in the early universe, vary significantly within the class of standard models…

The B-modes of polarization have not yet been seen, which is consistent with many, though not all, of the standard models.

About the IS&L “unexpectedly small (few percent) deviation from perfect scale invariance” all GKL&N have to say is

The standard inflationary models… predict the statistical properties of the faint ripples that we detect in the cosmic microwave background (CMB). First, the ripples should be nearly “scale-invariant”

This doesn’t seem to address at all the IS&L claims, which they make in more detail as

The latest Planck data show that the deviation from perfect scale invariance is tiny, only a few percent, and that the average temperature variation across all spots is roughly 0.01 percent. Proponents of inflation often emphasize that it is possible to produce a pattern with these properties. Yet such statements leave out a key point: inflation allows many other patterns of hot and cold spots that are not nearly scale-invariant and that typically have a temperature variation much greater than the observed value. In other words, scale invariance is possible but so is a large deviation from scale invariance and everything in between, depending on the details of the inflationary energy density one assumes. Thus, the arrangement Planck saw cannot be taken as confirmation of inflation.

GKL&N argue for three other confirmed predictions of inflationary models:

Second, the ripples should be “adiabatic,” meaning that the perturbations are the same in all components: the ordinary matter, radiation and dark matter all fluctuate together. Third, they should be “Gaussian,” which is a statement about the statistical patterns of relatively bright and dark regions. Fourth and finally, the models also make predictions for the patterns of polarization in the CMB, which can be divided into two classes, called E-modes and B-modes. The predictions for the E-modes are very similar for all standard inflationary models

On these issues I don’t see anything from IS&L and would love to hear from an expert.

The main issue here comes down to the question of the flexibility vs. rigidity of inflationary models. Is the inflationary paradigm rigid enough to make solid predictions, or so flexible that it can accommodate any experimental result? GKL&N are making the case for the former, IS&L for the latter, and they point out the following quote from Guth himself:

when asked via email if they could name any pro-inflation scientists who believe that the theory is nonetheless untestable, the trio pointed to a video of a 2014 panel during which Loeb asks Guth directly whether it’s possible to do an experiment that would falsify inflation.

“Well, I think inflation is a little too flexible an idea for that to make sense,” Guth replied.

A fair take on all this would be to note that it’s a complicated situation, and I doubt I’m the only one who would like to see an even-handed technical discussion of exactly what the “simplest” models are and a comparison of their predictions with the data. Claims to the public from one group of experts that Planck data says one thing, from others claiming it says the opposite are generating confusion here rather than clarity about the science.

I’m strongly on the side of IS&L on one issue, that of the danger of theories that invoke the multiverse as untestable explanation. I don’t think though that they make a central issue clear. The simple inflationary models whose “predictions” for Planck data are being discussed involve a single inflaton field, with no understanding of how this is supposed to couple to the rest of physics. One is told that eternal inflation implies a multiverse with different physics in different universes, but in a single inflaton model this physics should just depend on a single parameter, and such a theory should be highly predictive (once you know one mass, all others are determined). What’s really going on is that there is no connection at all between the simple single field models that GKL&N and IS&L are arguing about, and the widely promoted completely unpredictive string theory landscape models (involving large numbers of inflaton-type fields with dynamics that is not understood).

I think IS&L made a mistake by not pointing this out, and that Guth, Linde, Nomura and some of the signers of their letter (e.g. Carroll, Hawking, Susskind, Vilenkin) have long been guilty of promoting the defeatist pseudo-scientific idea that “evidence for inflation is evidence for a multiverse with different physics in each universe, explaining why we can’t ever calculate SM parameters”. By defending the predictivity of “inflation” while ignoring the “different physics in different parts of the multiverse” question, I think many signers of the GKL&N letter were missing a good opportunity to make common cause with IS&L on defending their science against an ongoing attack from some of their fellow signatories.

Update: There are sources with technical details of the arguments being made by both parties:

I’ll try to find time to read these carefully and try and understand exactly what claims are being made. Would love to hear from experts who may have looked at these and are better placed to evaluate what the arguments are.

This controversy continues to involve an unusual level of PR rather than science. The Stanford press office has just put out this, where Linde makes it clear that he sees this as a political and PR fight:

Linde added that he worries about the younger generation of scientists getting the wrong impression from this story. “I don’t want them to read this article and think that they are spending their time on inflationary theory in vain. But the enthusiastic support that we are receiving makes us optimistic that this is not going to happen,” he said.

There’s no mention in this press office story of their last press office story about Linde and inflation, which promoted the BICEP2 “smoking gun” vindication of Linde and inflation.

Some more takes on this story can be found here, here and here.

Update: Another article about this is at the Atlantic. A crucial issue here is whether inflation has now entered the realm of unfalsifiability. Given any likely new data that could appear, is there any way it could falsify inflation, or can one just come up with some version of inflation that will match it? Guth and Linde seem at times to be taking the attitude that this is fine, I take Steinhardt et al. as pointing out that this is no longer conventional science. Replacing falsifiability by arguments about how many prominent people have signed your letter is a worrisome development. From the article:

In 2014, for example, Loeb asked Guth during a panel discussion if inflation was falsifiable—whether you could design an experiment to disprove it. Guth called that a “silly question,” suggesting that inflation was an umbrella over many theories, making it very hard to knock them all out at once. The hope right now, he says, is to use observation and further theory to winnow inflation down to just one specific version.

“Our point is that this kind of reasoning is inconsistent with normal science and cannot be resolved by invoking authority,” Ijjas, Steinhardt, and Loeb wrote to The Atlantic.

: More about this from John Horgan.

Update: It seems that Andrei Linde is a Lubos Motl fan. This is getting very weird. It’s not normal to respond to a scientific argument by enlisting letter writers on your behalf, even less normal to put your university press office to work on a response, and truly far out there to think that it’s helpful to have Lubos announce that you have eaten from the tree of knowledge and that your opponents are imbeciles.

Update: For a sensible take on this that I think likely reflects well the views of most mainstream cosmologists, see Peter Coles.

Update: IJS have put together a Fact-Checking page. It lists the four “predictions” of inflation claimed to agree with experiment by Linde et al. and gives four references to papers published by Linde touting different “predictions” for the same quantities, predictions not agreeing with experiment.

This month’s Scientific American has a bizarre cover story by Nomura on “The Quantum Multiverse”, see here.

All in all, I don’t know what other people’s reactions to this have been, but before this started I was a lot more sympathetic to the argument that Steinhardt was treating the case for inflation unfairly.

Update: Yet more coverage trying to make sense of this, from Nick Stockton at Wired.

This entry was posted in Favorite Old Posts, Multiverse Mania. Bookmark the permalink.

53 Responses to A Cosmic Controversy

  1. John says:

    I found the following talk by Steinhardt, where he makes arguments similar to those in the scientific american article, particularly clear:

  2. Lino D'Ischia says:

    I thought this was bit ironic:
    GKL&N write:
    “Some of the standard inflationary models have now been ruled out by precise empirical data, and this is part of the desirable process of using observation to thin out the set of viable models. ”

    It would seem that the multiverse puts an end to “using observation to thin out the set of viable models.”

  3. Ilyas says:

    I think it might also be worth pointing out the value (yet again) of Scientific American in providing a great platform for disparate views and giving them air time. These articles remind us that just because there might be a “scientific orthodoxy” does not mean it is correct. The challenge to String Theory was carried and written about very well in SA who refused to be pushed about by the orthodoxy who challenged the challenger (e.g. people such as you) by bringing together many “prominent” scientists with leading faculty positions, and effectively saying “you cannot be correct since look at all these people who occupy top positions and who support string theory – they cannot be wrong”. We should applaud SA for continuing to endorse its fine investigative tradition.

  4. Another Anon says:

    Yes, I agree with Ilyas. Publishing an article about orthodox inflation is not an interesting thing for Scientific American to do. It’s like criticising a newspaper for not printing a “nothing much happened today” on a quiet news day. The story is the challenge to the orthodoxy.

    And I know that one of the signers is privately dubious of inflation. If a bunch of big names ring you up and ask you to sign their petition, you’re going to do it.

  5. hyd says:

    Even if all the people of this generation and the preceding generations were content with the idea that the SM parameters be inexplicable but a result of playing dice, I’m sure there would be people in the next generation and the next next generation … insatiable with that idea and looking to inquire further, as history has repeatedly told us about human’s inextinguishable sense of curiosity. The attempt of silencing heretics by a bunch of big names is as amusing as ‘the 100 scientists pronouncing Einstein’s theory of relativity is wrong’.

  6. tulpoeid says:

    This wild attack from the side of theistic theories makes me sad. Statements like “the B-modes of polarization have not yet been seen, which is consistent with many, though not all, of the standard models” fill me with wrath. Keep up the good work and congratulations for putting names such as Hawking’s in the same sentence with pseudo-scientific.

  7. Jonatan says:

    I’m not a cosomologist but I’ve followed the debate between Steinhardt et al. and the pro-inflation side for a while.

    To me, the lack-of-prediction issue raised by Steinhardt is very simple, fundamental and philosophical in nature.

    As I understant it, the situation is thus: inflation provides a mathematical model/framework that CAN produce a universe like the one we see with respect to certain observables (geometry of spacetime, statistical distribution of CMB anisotropies, etc…). These are the “predictions” celebrated by GKL&N. However, quite generically (when considering quantum effects), it also produces all other possible universes with respect to these observables (the multiverse/multimess).

    Therefore it predicts everything/nothing (sounds familiar ?). How could the fact that we observe a flat, isotropic, etc universe be counted as evidence for inflation if we could have just as well observed a curved or open or anisotropic one according to inflation ? Why is this obvious and deep problem not more openly acknowledge by the prononents of inflation is beyond me.

    Steinhardt has provided technical explanations elsewhere for why this is the conclusion one should reach about the state of the inflationary paradigm at the moment, and those explanations seemed quite convincing to me. I haven’t heard any convincing counterargument to this general conclusion yet. Basically: the very concept of prediction becomes meaningless in the context of a bubbling multiverse. Peter is of course quite familiar with this epistemological fact.

    The second problem Steinhardt raises is equally damning for inflation. The fine-tuning of the initial conditions for the parameters of the inflaton field (a problem acknowledged by Sean Carroll in his blog entry on the subject) seems to be equal if not higher than the one required by standard big bang cosmology to produce a flat, isotropic universe like the one we see. I think Penrose have also mentionned that point in the past. So basically, it seems that you’re left with a choice of fine-tuning the initial conditions of standard big bang cosmology, or fine-tuning the initial conditions of a yet unknown, unseen, “inflaton” field…

    If true, this is highly ironic considering inflation was invented precisely to avoid this fine-tuning of initial conditions in standard big bang cosmology.

  8. Ben says:

    I think one of the problems is that there’s “inflation” – the idea of an early period of exponential expansion driven by something like an effective scalar field, and then there’s “inflation” – a more complete description of where that field comes from and how we got into the inflating state to begin with.

    The former seems to make testable (but maybe not unique) predictions, such as near-homogeneity, adiabatic perturbations, etc. The latter seems to screw all that up with goofy things like multiversal bubble nucleation, eternal inflation, and so on.

    So some (Steinhardt et al) argue that since attempts to come up with a complete picture of inflation wind up being non-predictive, that the whole thing is intellectually bust. Others (some of the inflation theorists) disagree on “predictivity” grounds and get into arguments about multiverse stuff. Still others (most experimentalists) would rather stay out of the weird foundational stuff, focus on “inflation” as an incomplete model for a specific epoch of history, and test the “predictions” that way.

    I think it’s hard to call the experimentalists wrong, since there’s basically no complete model of the early universe, inflationary or otherwise. Even Steinhardt is pretty clear that his pet ekpyrotic scenario has serious gaps.

  9. Shah-hid says:

    Here Dr. Neil Turok gives an epic summary of inflation’s curiosities:


  10. Pasa Dina says:

    Thanks for all the relevant links (as usual), Peter.
    The Forbes article – sigh.
    The author totally misses what the debate is about. He thinks it is about the variety of possible potentials rather than the initial conditions or multiverse. And he asserts that everyone agrees there had to be rapid expansion, apparently missing entirely the bounce.

  11. cosmology grad student says:

    “One is told that eternal inflation implies a multiverse with different physics in different universes, but in a single inflaton model this physics should just depend on a single parameter, and such a theory should be highly predictive (once you know one mass, all others are determined).”

    This is a key point. For a single-field slow-roll model, eternal inflation affects predictions of e.g. things you can observe in the CMB only in that instead of one particular distribution of classical spacetime geometries after inflation (drawn from a power spectrum with one particular value of e.g. spatial curvature, spectral index, “average temperature variation” = delta rho/rho, …) there is another, different distribution (instead of drawing from a fixed power spectrum at the end of inflation, there is some probability distribution of power spectra).

  12. neil says:

    I read somewhere, possibly here, that the strength of a theory depends as much on what it excludes as what it explains. In this debate, I think Guth et al focus on what they believe inflation theory explains whereas Ijjas et al focus on what it excludes (nothing?)

  13. skydivephil says:

    I am the producer of a series of films on early universe cosmology and have interviewed Guth, Nomura and many other supporters, and critics of inflation ( including those that promote others models such as CCC, string gas cosmology and the pre big bang).

    I am not a cosmologist but someone interested in cosmology and have spent many hours discussing these issues with some of the leading people in the field.
    In our film on eternal inflation we interview George Efstathiou who gave the press conference for the cosmology results for Planck.
    Efstathiou claims that the Planck data favour models of inflation that are eternal and therefore a multiverse.
    you can see that 12:41 into our film here:

    Whats interesting is that in the ISL paper they make the same claim:
    “The plateau-like potentials selected by Planck2013 are in the class of eternally inflating models,”
    We asked Guth and Nomura to respond to many of the criticisms of the ISL paper and so I hope you will find our film interesting. Nomura claims a method of how to falsify eternal inflation.
    What I also find interesting is that Steinhart in his paper “Cosmological Perturbations: Myths and Facts” said
    “We will argue below that inflation and the cyclic model make firm predictions for the characteristic mass scale and equation of state during inflation, the spectral tilt of the scalar perturbations and the gravitational wave amplitude.”
    He also states that Ns should be roughly equal to .95 in inflation , Planck measured .96 as far Im aware.
    All of this was written well after it was realised that inflation led to a multiverse.
    We also discuss other ways to more directly probe for the multiverse ( and not just bubble collisions which we also discuss) and in other films how to probe for alternatives to inflation.

    My recollection of Steinhardt and Turok’s book “Endless Universe” is that inflation makes 6 predictions and 5 have been confirmed by experiment but the last hasn’t and they predicted it would fail at the last hurdle. Hence this justifies the notion of smoking gun. But they accepted it had past the first 5. This is the prediction B mode polarisation and primordial gravitational waves. In other words one could experimentally discriminate between inflation and the Ekpyrotic cyclic model by this signal assuming its from gravitational degrees of freedom in the early universe and not primordial magnetic fields, hence a detection of this signal would rule out this cyclic model and others cyclic models ( Penrose’s CCC for example ).

    It would not rule out all competitors to inflation though. String gas cosmology would still be in the game. But string gas cosmology ( according to Ali Nayeri who we interviewed ) predicts a red tilt for the power spectrum and blue tilt for the gravitational wave spectrum whereas inflation predicts both should be red tilted. There may be an exception in some inflationary models but then they would predict other non guassianties. So the picture is more complicated than the smoking gun headlines implied but I believe it has some truth. In others words different inflationary does may predict different strength of the signal but a common feature is its tilt.

    I think its clear inflation has built up a large body of evidence and is the leading horse in the race to describe the early universe. However it hasn’t crossed the finishing line yet and until then it its well worth exploring other alternatives. But these other alternatives are usually cyclic models which in some sense implies a world ensemble separated in time rather than in inflating space. It seems pretty much no one in this debate is arguing for a single universe from one big bang, that is a fact. Both sides that you quote in this controversy say there were multiple big bangs; the debate is, are they separated in time or space? This is something that I think is a very profound feature of this debate.

  14. Jonatan says:

    Do proponents of eternal inflation assert that all bubble/pocket universes of the multiverse are flat/isotropic/scale invariant/adiabatic etc… (I guess not since Guth says that “anything that can physically happen will happen an infinite number of times”)

    Because if they don’t, then I don’t see how observing any of this in this universe can count as evidence of eternal inflation. After all, someone standing in another pocket universe that is not flat/isotropic/scale invariant/adiabatic (and there would be an infinite number of those) would then be forced to conclude that eternal inflation cannot be correct if these features are true predictions of eternal inflation…

  15. Peter Woit says:

    I’m not competent or willing to moderate a general discussion of cosmology here. If you have knowledgeable comments specifically about the topic of the posting, please contribute them, I’m going to be deleting others.

  16. Shantanu says:

    Peter let me make a couple of sociological points. Its again sad that none of these articles menion (or interview) Demos Kazanas even though his 1980 paper says in plain English that accelerated expansion could solve the horizon problem. Also its interesting that there are people from Harvard on both sides of this debate
    cosmology grad student : I think given any observations you can concoct any model of inflation. There are even models of open inflation (despite the claim that inflation predicts \Omega=1)

  17. The key point that is underappreciated in discussions here is the one that Guth et al. make in their article, where it says.

    They contend, for example, that inflation is untestable because its predictions can be changed by varying the shape of the inflationary energy density curve or the initial conditions. But the testability of a theory in no way requires that all its predictions be independent of the choice of parameters. If such parameter independence were required, then we would also have to question the status of the Standard Model, with its empirically determined particle content and 19 or more empirically determined parameters.

    Indeed, this is how physical theories make predictions generally: first to fix some of the parameters to specify a model in the theory (a solution), then the remaining properties of that model are its predictions.

    If one took the standpoint that the theory as opposed to its models is to make fairly unique predictions as such, then none of the usual theories would fit.

    The usual established theories (as opposed to the models built inside them) don”t even have large finite (say of cardinality 10^500) but have usually hugely non-finite-dimensional spaces of solutions.

    Those readers who really think this is controversial need to pause for a moment to think about this elementary fact.

    For instance the solution space to Einstein gravity, unconstrained, is humongously infinite dimensional and says nothing specific about the observable world, it doesn’t make any predictions whatsoever — not before we pick in there a very low-dimensional parameter space of models, say the 3-dimensional subspace of simple FRW models. We pick such subspaces not for some a priori reason, but because we match these three parameters to experimental observation. The remaining properties of the FRW models within the theory of Einstein gravity, that is the predictions of this model.

    Similarly quantum field theory as such makes virtually no predictions, not before some model is specified. The best that QFT by itself can say is that the universe is described by some local Lagrangian density on some space of fields such that quantum anomalies cancel. But the moduli space of all possible field species and all possible local Lagrangians on them, that’s a humongously infinite-dimensional space. No prediction is to be derived from that alone. Predictions are only being derivable once loads of parameters are specified: First we specify (by hand, because it fits observation) the local Lagrangian density to be of Einstein-Yang-Mills-Dirac-Higgs-type, then we specify the exact particle species, then their Yukawa couplings and so on. Only once all these parameters have been fixed by hand does the theory start making predictions.

    With the theory of cosmic inflation it is exactly the same.

    This doesn’t imply that cosmic inflation is right. But this means that if you think it is wrong for this reason, then you would have to regard every established physical theory as wrong, too. Because they all share this feature. A physical theory does not make predictions before models have been picked inside it, and they all admit alternative models that differ drastically from the one you want to pick because it fits observation.

  18. Peter Woit says:

    You continually make this claim that two completely different things are “exactly the same”, I’ll just point out that this FAQ applies here


  19. Jonatan says:


    Steinhardt’s point, unless I missunderstand him, is that generically, in an eternally inflating multiverse, even once you’ve chosen a specific inflaton potential and parameters, because of quantum effects, the values of observables such as spacetime geometry and the statistical distribution of CMB anisotropies, will take any possible values in the ensemble of pocket universes.

    If that’s correct, than how can specific values for these observables in our universe be considered “predictions” of eternal inflation as they usually are ?

    What should people finding themselves in a pocket with open spacetime geometry or non-gaussian distributed anisotropies be expected to conclude about eternally inflating models ?

  20. spacetime geometry and the statistical distribution of CMB anisotropies, will take any possible values in the ensemble of pocket universes.

    If that’s correct, than how can specific values for these observables in our universe be considered “predictions” of eternal inflation as they usually are ?

    To repeat, this is the usual state of affairs as it has always been: We make some observations about the universe around us (“pocket universe”, if you really like to say that), fit a model of our theory to that set of data and then make predictions for all the remaining observations.

    This is no different from how we predict anything, say the weather. We first measure the mess that we find ourselves in, then you apply theory to that to see how it evolves.

    I find it bemusing that many participants here have effectively become strong Hegelians, demanding a derivation of the entire universe from pure thought alone, without experimental input, without model building, and complaining that available physical theories are not like this. They never have been.

    Maybe one day we will have such a theory, and Hegel will be vindicated. It’s a logical possibility that such a theory exists. But if it does, this will be cause of awe and amazement, not something to take for granted, while as long as it does not, everything is business as usual.

  21. Steinhardt et. al write here:

    The Planck satellite results […] are stunning. For the first time in more than 30 years, the simplest inflationary models, including those described in standard textbooks, are strongly disfavored by observations. Of course, theorists rapidly rushed to patch the inflationary picture but at the cost of making arcane models of inflationary energy and revealing yet further problems.

    In view of the remarkble likelyhood plot that PLANCK15 produced (fig 12 in PLANKC 15 XX “Constraints on Inflation”), this is an unexpected statement. Right there in the middle of the plot sits Starobinsky’s R^2 inflation, the first model for inflation ever proposed. The data does not significantly discriminate between Starobinsky inflation and other plateau-type inflationary models but it does prefer this class.

  22. Peter Woit says:

    Which of the following possible alternatives do you think is more likely?

    1. Ijjas, Steinhardt and Loeb are ignorant about basic facts of their subject like the one you are explaining here.
    2. They aren’t.

    Personally I’m inclined to 2, and in general to the assumption that both sides in this argument know a lot more about the subject than I do (and than you do). Given that, it might be best to pay close attention to the technical counter-arguments each is making when they discuss the same issue and see if one can figure out what the story is there, leave the accusations of incompetence and not knowing what they are talking about to Lubos.

  23. Marc Nardmann says:

    Urs Schreiber wrote: “But the moduli space of all possible field species and all possible local Lagrangians on them, that’s a humongously infinite-dimensional space. No prediction is to be derived from that alone. Predictions are only being derivable once loads of parameters are specified: First we specify (by hand, because it fits observation) the local Lagrangian density to be of Einstein-Yang-Mills-Dirac-Higgs-type, then we specify the exact particle species, then their Yukawa couplings and so on. Only once all these parameters have been fixed by hand does the theory start making predictions.”

    We have to specify only the particle field species (leptons, quarks, gauge bosons, Higgs), the symmetry group (spacetime symmetries and internal symmetries), and how the particle fields transform under those symmetries. (This input is essentially discrete, not “humongously infinite-dimensional”.) The Lagrangian density is then the most general function that is invariant under the group action; we do not have to put it in by hand. Its renormalisable part turns out to be determined by a small number of parameters (coupling constants, masses, …). Mathematically speaking, the Lagrangian arises from a theorem, not from a definition. This is considerably less arbitrary than it sounds in your description.

  24. Jonatan says:

    Urs Schreiber wrote: “We make some observations about the universe around us (“pocket universe”, if you really like to say that), fit a model of our theory to that set of data and then make predictions for all the remaining observations.”

    With all due respect, I don’t think you understand what the point is. As Peter said, I doubt that Steinhardt is that mistaken about the idea he co-invented.

    If the model that fit the “local” data turns out to also produce every other conceivable data set, how could you ever test whether this is the correct model ? How do you know whether what you observe is a common or very unusual case within the infinite multiverse ? This is the measure problem that Guth et al. recognize and is unresolved to this day. To repeat, what should someone living in a non-flat, non-isotropic, non-scale invariant part of the multiverse conclude about the status of inflationary cosmology ?

    Sure you’ve got a rich mathematical framework capable of modelling anything and everything, but what have you learned about our physical reality. In my view, precisely nothing.

    You seem to think that this is a common situation in science but as far I know, this unfortunate problem is entirely peculiar to both string theory and inflationary cosmology.

  25. Anonyrat says:

    IS&L write at one of the links in this blog post (and thus answer Urs Schreiber):

    What about the comparison to the Standard Model of Particle Physics? This comparision is a false equivalency. For the Standard Model, there are definite predictions for any choice of parameters. For Inflation, there is an infinite diversity of outcomes for any choice of parameters (i.e., for any choice of the inflationary energy curve). For example, for any one choice of parameters, an infinite number of patches of space in the multiverse are produced that are not flat, not smooth, and do not have the properties astronomers observe – and there is nothing in the inflationary theory to say that one outcome is more likely than the others. The same does not apply to the Standard Model of Particle Physics.

  26. Anonyrat says:

    Urs Schreiber wrote:

    In view of the remarkble likelyhood plot that PLANCK15 produced (fig 12 in PLANKC 15 XX “Constraints on Inflation”), this is an unexpected statement. Right there in the middle of the plot sits Starobinsky’s R^2 inflation, the first model for inflation ever proposed. The data does not significantly discriminate between Starobinsky inflation and other plateau-type inflationary models but it does prefer this class.

    Around minute 37 of his talk here http://vms.fnal.gov/asset/detail?recid=1944338 Steinhardt addresses this point. In brief, one must rule out all models with eternal inflation.

  27. Jonatan says:

    Indeed, the point I’m refering to is the one Steinhardt explains in the video Anonyrat mentionned, starting at 28:45 (the part around 35:30 is particularly relevant). He also talks about the measure problem starting around 37:50.

  28. srp says:

    I would like Urs to either respond to Anonyrat’s quotation or admit that he was wrong in his insistence that inflation models are predictive in the same sense as the SM. It directly contradicts his claim that once you put in the parameters for an inflation model you get an unambiguous prediction of everything else.

  29. qwer1304 says:

    It’d seem that the apparent necessary entailment of multiverse (multimess) is the strongest objection to inflation (as Steinhardt says himself). In that light, it’d be interesting to hear both parties’ thoughts about V.Mukhanov’s (who’s actually the first who worked out quantum fluctuations source of matter formation in 1981 http://www.jetpletters.ac.ru/ps/1510/article_23079.pdf) model without self-replication https://arxiv.org/pdf/1409.2335.

  30. Anonyrat says:

    I’ve been trying to come up with the most sympathetic view that I can of postmodern science. The best I can do is as follows:

    The framework of classical Hamiltonian mechanics is of interest because specific Hamiltonians are highly relevant to systems we observe or build. The framework of the Schrodinger equation is relevant because specific Hamiltonians are highly relevant to the real world. The framework of Quantum Field Theory is likewise useful, because specific field theories have proven to be highly predictive.

    In postmodern science, it seems that the paradigm is that the framework will give rise to a myriad of scenarios, of which the scenarios relevant to our world are few and even relatively unlikely to arise. One might tolerate this if these postmodern frameworks could say – conditional on your pocket universe having these handful of properties, here is a plethora of other things the framework predicts for your pocket universe with no wiggle room.

    The problem with the postmodern framework of inflation is that there is no plethora of predictions – as I see it, there are only a handful – and the wiggle room remains immense. The problem with the postmodern framework of string theory is that even with everything we know about physics – let alone a handful of facts – as given properties, there are no testable predictions; our sum total of knowledge is insufficient to unambiguously determine the string vacuum – there are no predictions, there remains a lot of wiggle room. However, these two failures do not mean that a tolerable postmodern theory cannot exist.

  31. Anonyrat says:

    To complete the above thought with a metaphor: a tolerable postmodern theory of physics must produce a “discrete spectrum” of universes, not a “continuous or near-continuous spectrum”, otherwise there is always wiggle room, and one would have to stipulate an infinity of facts about our universe to fix our position in the “spectrum”.

  32. Peter Woit says:

    I don’t think discrete vs. continuous is relevant. What is relevant is the explanatory power of your theory: a good theory makes lots of non-trivial testable predictions based on a small set of assumptions or choices. A bad theory is one where for each testable prediction you need a new choice or assumption. The argument here is that inflation has moved to the bad theory side: Linde is quite explicit at times that what he is doing is developing versions of inflation that can match any new observation of B-modes, of non-gaussianity, etc. This was the question Guth was being challenged on and I don’t think had an answer for: is “inflation” now completely insulated from falsification, and if so, is it still science?

    I don’t think using the term “postmodern physics” is a great idea unless it gets a well-defined meaning, otherwise it will just get thrown around as a sort of meaningless insult (or maybe, for some people, a badge of honor…)

  33. Jonatan says:

    It seems to me that one common feature of both string theory and inflationary cosomology is that in both cases, a new “piece of furniture” in physical ontology was invented (the string and the inflaton field) not because they were required to make sense of new observations that couldn’t be explained by the existing physical objects making up the ontology of current physical theories, but rather to improve on existing theories by way of unification, the resolution of fine-tuning issues, etc…

    I wonder if this kind of liberty (one might say self-indulgence) on the part of theorists is not the common “cardinal sin” responsible for the overabundance of degrees of freedom leading to multiverses and rendering both of these ideas ultimately meaningless scientificallly speaking.

  34. Peter Woit says:


    An important aspect of the inflation story is that it’s roots are in the study of the Higgs field, and more general uses of scalar fields to make viable GUT theories. What got people excited about the idea was not just the supposed explanations of otherwise hard to understand general aspects of cosmology, but the fact that these were coming from postulating not some random new thing, but something already being used for other promising purposes. I think Guth’s original dream was that the same scalar field would break GUT symmetry and provide inflation, giving two completely different sorts of evidence for it, which would be very compelling.

    The “cardinal sin” of introducing new degrees of freedom unrelated to anything else to explain something occurred later, as GUTs failed to work out (there are still proposals to revive connection to known physics, identifying the inflaton with the Higgs). ILS essentially claim that this has gotten worse in recent years, with the simplest models not working out, requiring what they call “postmodern inflation”. One troubling part of research into inflationary models now involves introducing large numbers of new degrees of freedom, killing the explanatory power of simple models, and yes, that’s properly something “sinful”.

    The story of string theory is unfortunately similar, as the models of the mid 1980s were understood to be problematic, and evolved into the absurdly more complex string landscape ones.

  35. Jonatan says:

    Thanks Peter for reminding me about the interesting history of the field. You’re right that the initial connection between the use of scalar fields in GUT and the development of inflation shows that the idea was not without broader motivations at the beggining.

    What always seemed suspect to me though was the ease with which fundamental physics since the early 1980’s turned to claims about the existence of new ontological entities (strings, inflaton) without any direct empirical reasons for doing so.

    I mean, the intelectual path do seem different to me than the way inferences for the existence of the rest of physical ontology (particles of the standard model, four dimensional spacetime, dark energy and dark matter) were made. That is, these inferences appeared much more directly related to our empirical experience of the world. In fact required in order to make sense of this experience.

  36. Yatima says:

    This will run the whole weekend for sure. It’s on RT right now:


  37. Peter Woit says:

    Thanks, good to get a take on where Putin stands on this.

    I wonder if the people who signed on to this letter thought much about what they were doing, whether backing Linde in a high-profile media fight was really a good way to defend the credibility of this field of physics.

  38. Jonatan says:

    Briefly restated, my impression is that theories are probably much more likely to end up in a “multimess” when they are based on an ontology invoked for theoretical convenience/ambitions rather than by empirical necessity.

  39. Art says:

    The original ISL critique (1304.2785) identified 3 issues, and spent the least number of words on the third, the eternal inflation implied by the data-favored plateau potentials. The specific criticism here was that the agreement of the Planck data with the nominal expectations of the models was “surprising” given infinitely many opportunities for large deviations. GKN (1312.7619) responded that there’s no reason to think that large deviations are probable.

    By contrast, subsequent discussion is heavily weighted towards the multiverse/multimess/multi-whatever. In my view, it’s deviated substantially from the original topic.

  40. Bill Kuncik says:

    As a layman very interested in science, I think this debate about inflation has done a good job airing and evaluating the subject issues. But I have to tell you, the manner in which it is being conducted is, least in my opinion, creating a bad public impression of the physics community.

    By my observation, from BICEP, to the stage of the World Science Festival in New York, to this latest exchange in Scientific American, and much in between, the expositions, pronouncements and exchanges have simply gotten too acrimonious, too personal, and too characterized by incendiary phrasings, media grandstanding, insufficient respect for the other’s position, and perhaps even for the other. By both sides, and to an extent one does not ordinarily see in such matters.

    Not by everyone, to be sure. But by enough to potentially discredit the enterprise of physics in the eyes of the public, which ultimately decides whether and how much to fund that enterprise.

    As was the case with BICEP, this latest episode seems to be getting legs in the general media. So while everyone understands that you all are human too, and that science is political like anything else, please press the reset button and try harder to stick to the issues. For the sake of

  41. a 1 says:

    ”we question whether the consequence of an incomplete theory
    can be legitimately called a ‘prediction’

    That’s from a really nice paper about infinitistic arguments in today’s physics:
    1604.06773 ”The infinite turn and speculative explanations in cosmology” by Reza Tavakol and Fabio Gironi

  42. Jonas says:

    As mentioned above Penrose has argued that inflation would just make the fine tuning problems for the initial stage of the big bang even worse.
    He gives a semi popular description of these ideas in this lecture

  43. Pingback: Is a Popular Theory of Cosmic Creation Pseudoscience? | MEDIAVOR

  44. Dave G. says:

    I just noticed that Ijjas et al. have put up a new page on their website that they call “FACT CHECKING.”
    It took me a while to wade through everything. But the page provides multiple examples of papers that contradict the claim that inflation has made predictions in the past that turned out to be successful.
    For each prediction, there’s a list of papers that claimed that inflation could also produce the opposite result. And it looks like those papers were all written before the observations were actually made.
    To make matters worse — the authors of the papers include Guth, Linde, Kaiser and Hawking.
    It doesn’t seem fair to claim that inflation predicted something if you also showed that the opposite was possible – you can’t just forget those earlier papers.

  45. skydivephil says:

    Peter , Im concerned about the use of the video where Guth says inflation is not falsifiable. I have only been able to find a very short version of the discussion and no context. do you have the whole discussion available? if so would very much like to see it.
    Im primarily asking because Loeb et l’s critique is more focused on eternal inflation than inflation per se and Guth and Nomura have published a paper showing how eternal inflation could be falsified. This paper is discussed by Nomura in our video about that 49 56 seconds . https://www.youtube.com/watch?v=QqjsZEZMR7I
    the link to the paper is here:
    Its true that there are version of inflation that are not eternal but according to Guth they are contrived and according to Efstahtiou they are not favoured by the data.

  46. Pingback: This Month’s Hype | Not Even Wrong

  47. Peter Woit says:

    The video where Guth says that is embedded in the Undark piece and you can there see it in context. I think it’s just the usual “you can’t falsify a framework, only models” argument that Urs Schreiber was making above.

    For the Guth/Nomura article, there’s a new blog posting…

  48. skydivephil says:

    Thank you Peter, but the video in the Unmark piece is only 40 seconds long. What I was looking for is the whole discussion. If you have a link to that, would be very much appreciated.

  49. Peter Woit says:

    There’s a link at undark, it’s to this

  50. skydivephil says:

    Thank you for that Peter look forward to watching it.

Comments are closed.