Multiverses: Science or Science Fiction?

The September issue of Astronomy magazine is now out, with a cover story on Multiverses: Science or Science Fiction? The author Bob Berman does a good job of explaining both the arguments for various Multiverses, as well as the reasons for skepticism about some of these arguments. After quoting Max Tegmark as defending multiverse theory as science since it is a prediction of an “arguably testable” theory (inflation), Berman ends the piece in a way I have to agree with:

Given the current multiverse infatuation, it may be fairest to give the last word to a prominent skeptic. Columbia University mathematical physicist Peter Woit, who maintains the popular multiverse-critical blog Not Even Wrong, pulls no punches.

“Physicists had huge success in coming up with powerful compelling fundamental theories during the 20th century,” he explains, “but the last 40 years or so have been difficult, with little progress. Unfortunately, some prominent theorists have now basically given up and decided to take an easy way out. The multiverse is invoked as an all-purpose, untestable excuse. They allow theoretical ideas like string theory that have turned out to be empty and consistent with anything to be kept alive instead of abandoned. It’s a depressing possibility that this is where physics ends up. But I still hope this is a fad that will soon die out. Finding a better, deeper understanding of the laws of physics is incredibly challenging, but it’s within our capability as humans, as long as the effort is not overwhelmed by those selling a non-answer to the problem.”

Whoa, intense. We’ve got to toss the multiverse if we care about physics!

Of course, if an infinite multiverse does exist, some other Woit is out there saying the exact opposite.

The same issue of Astronomy has a “Web Extra” entitled What happens if string theory is wrong? It mentions the 2013 poll of theorists discussed here, which had a large majority (73%) answering the question “Do you think that String Theory will eventually be the ultimate unified theory?” with a “No”, then goes on to link to a 2007 article by Sten Odenwald. Some of that article includes quotes from an interview with Lenny Susskind, which Odenwald recently included here. It will be interesting to have an update on that material in a year or so once 13 TeV LHC results on supersymmetry are in.

Bonus material: Quanta magazine has a great interactive map of “Theories of everything”.

This entry was posted in Multiverse Mania. Bookmark the permalink.

59 Responses to Multiverses: Science or Science Fiction?

  1. Bernhard says:

    Does Tegmark mention why inflation predicts a multiverse? I suppose the argument is that some models of inflation predict a multiverse (I’m not even sure what that really means, but anyway) ? In the case of string theory is clear what happens, the theory predicts nothing, so the multiverse is used as an excuse for this failure and it also become the main prediction of the theory. But inflation is supposed to make a few predictions (right?). I’m sure one can have inflation without a multiverse, so the whole connection is rather unclear to me. As I’m ignorant in this whole thing I just looked at Wikipedia, where I read “All models of eternal inflation produce an infinite multiverse, typically a fractal.” Whatever that means.

  2. Pingback: Multiverses: Science or Science Fiction? | Not Even Wrong - SciFi Picks

  3. Peter Woit says:

    I’m no expert, but from what I’ve seen of work on inflationary models, Steinhardt is right that you can get pretty much whatever you want from them, and I assume that means a multiverse or no multiverse. One expert is Katie Freese, and here
    she comments on the BICEP2 data (which turned out to be wrong)
    “If you take the BICEP data literally, which I’m not saying you should, you never have eternal inflation. So you don’t have to have eternal inflation, if you ask me. I was very happy about that.”
    See her paper with Will Kinney about this
    where the abstract says
    “even tiny running of the scalar spectral index is suffcient to prevent eternal inflation from occurring…”
    I have seen multiverse enthusiasts claiming “inflation implies multiverse”, but since many of their other arguments are full of holes, I see no reason to believe this one isn’t too.

  4. Curious Mayhem says:

    It’s still hard to understand why people try to connect “multiverses” (in the string sense) with inflation. The two logically have nothing to do with each other. Inflation can give rise to different parts of spacetime having different elementary physics or “eternal” inflation, but they’re all part of one meta-universe, or whatever you want to call it. And neither has anything to do with the many-worlds interpretation of QM, which is not a theory, but a particular spin on quantum mechanics, if you’ll pardon the pun. (Arguably, it’s an obsolete interpretation, superseded by a better approach, decoherence, which is actually testable.)

    Inflation could be proved tomorrow and have no direct bearing at all on string theory. There is an indirect bearing: string theory has a very hard time accommodating inflation, except for some special models with a peculiar symmetry.

    (Interesting lookback: I just saw the old “East/West Coast metric” post. The history of the West Coast metric is simple: Bjorken & Drell’s 1964 textbook, before Feynman ever arrived at SLAC. The convenience for particle physicists is straightforward. Time-like and causal spacetime intervals have a positive square-interval, and most importantly, the momentum 4-vector has positive norm = +m^2. In a particle’s rest frame, this convention fits naturally with just a time-like component being non-zero. And the phase of a positive wave starts with +omega*t (or E*t in natural units), again matching nicely with the rest frame.

    (The East Coast metric does have advantages if you’re working with special or general relativity and need to make contact with the non-relativistic limit. But particle physicists mainly live in momentum space, not real space. Which might explain something about them ….)

  5. Peter Woit says:

    Curious Mayhem,
    I learned QFT from Bjorken and Drell, my invocation of Feynman was just a guess as to where they got their choice of metric from (they were writing in the early sixties, at a time there were already quite a few QFT books around). After I later checked a lot more QFT books I realized that many of the 1950s-era ones, for instance Schweber, or Bogoliubov and Shirkov, were already using the West Coast metric long before Bjorken and Drell. I’m still curious about the history, never managed to find out who first started using the West Coast convention. Interestingly, the earliest sources I looked at treated special relativity by using imaginary time (and thus the East Coast metric), which is very natural from the point of view of modern Euclidean methods.

    I don’t want to start up that discussion again, but would be interested if someone knew the 30s-50s history of this.

  6. Robert Delbourgo says:

    We were brought up on the West Coast metric at Imperial College in the late fifties/early sixties but that may have been the influence of Schweber’s book. Bogoliubov and Shirkov was also out then. Again that may also be due to the fact that we were being instructed in particle physics; general relativity took more of a back seat.

  7. Low Math, Meekly Interacting says:

    I’ll never get over my bafflement at the attraction to the anthropic multiverse. It’s not so much wrong as a decision to commit suicide, from a scientific perspective. What’s the point of a prediction that can never be tested? Even if one believed in the multiverse utterly, it’s still a trust fall into the abyss. Is it really worth redefining science to keep? Isn’t human dishonesty enough of a problem for science without having to dispense with the means to combat it?

  8. unification says:

    Inflation could be proved tomorrow and have no direct bearing at all on string theory. There is an indirect bearing: string theory has a very hard time accommodating inflation, except for some special models with a peculiar symmetry.

    ” string theory has a very hard time accommodating inflation” why is this so? there is a field string cosmology

  9. Bee says:

    Did you mean 13 TeV?

  10. Peter Woit says:

    Thanks, fixed. I need to figure out some way to stop always making that mistake..

  11. MB says:

    “Of course, if an infinite multiverse does exist, some other Woit is out there saying the exact opposite.”

    Well that’s an entertaining notion at least.

  12. anon says:

    Curious Mayhem, I am similarly puzzled by the seeming widespread confusion of multiverses (or phase regions in a single inflationary space-time) and many-worlds, which are completely unrelated concepts. I am also puzzled by the oft-repeated claim among otherwise serious people that every phase and every history that is possible in an infinite (possibly eternally inflationary or not) space-time is therefore *mandatory*. Maybe I am missing some ergodicity assumptions that they never state, but to me that sounds like like saying that because infinity is a lot, therefore any infinite sequence of integers includes every integer (obviously false).

    As for decoherence, though, since most people who study decoherence for good reason don’t claim that it is an *interpretation* of Quantum Mechanics, I disagree that it can supersede any candidate interpretation of QM, including many-worlds (which I personally disbelieve, by the way) but this may be veering off-topic.

  13. Wayne says:


    I’ve always wondered why multiverse proponents always say that the
    laws of physics and the values of fundamental constants vary in each
    universe. If there is a multiverse (or if eternal inflation is
    correct), why must this be so? Why can’t each instance of inflation
    somewhere just produce the same laws and constants?

    Also, I don’t understand why string theorists say that string theory
    predicts a multiverse. Do they know how string theory describes the
    making of these different universes? As far as I know, they just have
    different solutions, but no mechanism by which these universes can get
    made. Or perhaps I’m completely wrong.

  14. Peter Woit says:

    Likewise, I’ve never understood the “anything that can happen will happen” argument, it seems to be based on an assumption that the underlying theory is mostly vacuous (other than providing some constraints on what can’t happen). In particular, in any sensible theory, no Peter Woit would be defending the multiverse…

    As usual in this kind of “string theory”, you can come up with “string vacua” that give what ever you want. I’m sure you can come up with a string inflation model that leads to a multiverse with each universe having the same physics, as well as models with each universe having different physics. The appeal of the second is that you can use it as an excuse for not being able to predict anything, the first has the defect of being predictive.

  15. Scott Church says:

    Bernhard/Curious Mayhem/anon,

    I’m no expert either, but I believe the confusion over the term “multiverse” is that it’s used in different ways. Eternal inflation leads to one in that each of the “bubbles” that have reheated from the larger inflating “fractal” space-time (or “metaverse”) are being called “universes.” This is what Tegmark referred to as a “Level I” multiverse (Many Worlds “parallel” universes he calls “Level II”). Eternal inflation, if real, would lead to a Level I multiverse.

    These days I see a lot of folks doing two things: First, they treat inflation as synonymous with eternal inflation, which it isn’t (it is true that most inflationary models are likely to be eternal, but not all are, and those that aren’t seem to get summarily dismissed). Second, they treat any inflationary multiverse as one based on the string landscape even though, as you said Curious Mayhem, the two logically have nothing to do with each other.

    It seems to me this is happening mostly because in the absence of a truly workable theory, the eternal inflation/string landscape show is the only way to generate the widely varying physical parameters/laws one needs to anthropically sidestep fine-tuning issues. Without the string landscape, inflation (eternal or otherwise) only gives us a universe/s with the same physical laws as our own, and that doesn’t do much to keep the fine-tuning demons at bay. A lot of people these days are terrified of fine-tuning… and whether they’ll admit it or not, for reasons that seem to be more metaphysical than scientific (hence, the ease with which they accept post-empirical “beauty” as more scientific than direct observational verification). Over the last 30 years or so this is the best they’ve been able to come up with and they’re sticking to it… because admitting defeat and facing the alternatives is just too damn scary. 🙂

  16. Magnema says:

    @ Scott: You’re off-by-one on your numbers there, with respect to Tegmark.

    Level 3 is QM MWI. A note on this: anon is correct in stating that decoherence is a phenomenon, not an interpretation. It’s when a thermal system interacts with a quantum one, which effectively wipes away the quantum one – but this occurs in both MWI and Copenhagen-style work.

    Level 4, known in the philosophical community as David Lewis’ Modal Realism, is “everything possible exists” (up to some possible limitations). This has the strength of working from more-or-less no axioms (and therefore is supremely “simple”), but has the weakness that it’s virtually impossible to get anything out of it. I think this is the epitome of “simple theories over understandable predictions,” and is much more grounded in aesthetic metaphysics (and plausibly epistemology) than anything physicists (with the exception, as usual, of Tegmark) want to deal with.

    Finally, there’s the issue of the multiverse as it’s most often discussed on this blog and similar sources, which Tegmark splits in half. You start out by realizing that if we have a really big universe, then there’s a lot of parts that we haven’t seen to date, and – given dark energy – we might never see, and therefore we can never observe. Inflation, in itself, gives you nothing more than what a really big universe gives you (and I believe this is true for both eternal and non-eternal theories). There’s the issue of how seriously to take these other possible universes, but if they’re all virtually identical then there isn’t much of an issue here. (This is the kind of multiverse which, in the most straightforward sense, a positive curvature would rule out.)

    The distinction here between “inflationary universe” and “infinitely large ordinary universe” only occurs if you add some theory – any theory – which has multiple low-energy limits, such as string theory. Then, you can get the anthropics on parameters, which is the whole point of people talking about the multiverse in the first place. This is where your numbering mismatch with Tegmark comes from – he calls the former, a really big universe, the level 1, and the latter, where you can get anthropics in physical laws, the level 2.

    @Wayne: The (simplified) idea is that different parts of our universe would have no causal contact, and therefore, while they could be the same, there’s no reason to be – and if there’s infinitely many causally-independent regions, then with probability 1 they will have a representative from every non-probability-zero subset.

  17. Magnema says:

    Insofar as predictions and testability are concerned: I think it would be fair to say that a single-universe perspective makes the prediction that life should be a reasonably generic property of possible universes, given that the one that exists has life (barring both incredible luck and any deific-style selector on life existence). Any multiverse makes no such prediction, although the absence is a prediction itself relative to a single-universe perspective.

    These predictions are highly unconventional in that, unlike typical science, these start with the experimental result (life being observed) and make conjectures about the theory, because they make guesses about the space of possible worlds based on our observed one. It’s analogous to a physicist making a mathematical conjecture based on a particular model of an observed physical phenomenon, and then (when mathematicians prove the physicist’s conjecture) taking it to be evidence that his model of the phenomenon is correct.

    It’s odd, to be certain, and brings to bear a question in the philosophy of science, namely: what are predictions allowed to be about? Are you allowed to conjecture something based on your theory which is logically either true or false before you conjecture it, then treat that as a prediction? In other words, should you count predicting math with physics as a testable prediction? Personally, I think it would be fair to say yes, but I’m willing to grant that it’s not clear…

    A mostly-related remark: I started following this blog because I thought it would be a good thing for me to have both pro-multiverse and anti-multiverse resources on my reading list. I’m glad I did that, because my positions have shifted significantly by the exposure to more perspectives… while I maintain a MWI-view of QM (with some QB sympathies), and maintain an acceptance of the multiverse, it’s led me to question my views a bit more, as I’ve seen some legitimate criticism (as opposed to the flawed counterarguments I’ve heard most often).

    In particular, whereas before I took the above test to be “single universe predicts no fine tuning” (which I still do) and “fine tuning exists” (which logically leads to the conclusion “single-universe has been falsified”), this blog has convinced me that we don’t understand how to apply anthropics (either in “what life is” or in the prior probability of worlds) well enough to conclude the latter point. I’m still inclined to sympathize, but every time people try to use anthropics and make wrong predictions, I become significantly less certain that they can get the logic correct when they already know the answers they should get…

    (Sorry for the longwinded double-comment, but I figured since I was talking about two different things I should split it.)

  18. Peter Woit says:

    I don’t see any argument at all from “causally-independent” to “with probability 1 they will have a representative from every non-probability-zero subset” (and I think this is the point anon was raising). Just because you know nothing at all about causally disconnected universes doesn’t mean there are an infinite number of them out there uniformly filling out some probability space.

    To actually say anything about these other universes you need a theory of where they came from (and no, I’m not convinced such a theory will only produce negative curvature). The problem with all discussions of the multiverse is that they’re empty of actual substance: no one has a viable parametrization of the space of possibilities or of the dynamics leading to these possibilities. What people do then is just make an empty statement about things not understood being equally likely, and their “multiverse theory” is indistinguishable from my “I have no idea what is going on” theory.

  19. Scott Church says:

    Magnema, thanks for the sanity check! I actually meant to preface that statement with “If memory serves me…” but inadvertently dropped it in my edits. Apparently, memory didn’t serve me all that well. 🙂 Anyway, thanks for all your other comments too… very informative and thought-provoking.

  20. adrian says:

    Dear Peter,
    I am not fan of taking a quote and deconstruct it, as it may loose the intended meaning, but what you wrote above caught my attention. Indeed, the quote in your article (did I misunderstand? is it you saying this?)

    ”They allow theoretical ideas like string theory that have turned out to be empty and consistent with anything to be kept alive instead of abandoned. ”

    Seems to suggest that you believe that

    –Sting Theory is an empty idea.
    –that String Theory is consistent with anything.

    Is it the case that you believe on these two statements above?
    Could you please qualify them? what do you mean by ‘anything’?
    what do you mean by ’empty’?
    Are you suggesting that all the work done in String Theory since 1969 is empty of content? Could you please explain? I can think of some counter-examples to you two statements above.

  21. Magnema says:

    @ Peter Woit: Statistically, I believe it is fair to say that, for some distribution, in an infinite number of trials, you will, with probability 1, get at least one point (even infinitely many points) every non-zero measure subset, correct? If the subset has total probability p, then the probability of missing it N times is (1-p)^N, which goes to zero as N goes to infinity, meaning you will almost certainly hit it eventually.

    Now, you can question three things that I can see. Firstly is whether or not they are randomly selected at all, which I take to be anon’s point. I think the most plausible counterargument would be “if not randomly, then what?” In general, I don’t see a way this is distinguished from any other statistical mechanical problem, where you have probabilities arising from a deterministic problem based on one’s own ignorance. Of course, there’s foundational questions in statistical mechanics, too, but at the very least I don’t think this is a different question.

    Secondly, there’s the more technical issue of where the probability measure comes from. I accept this as an issue, both theoretically and (as evidenced by the repeated failures of anthropics) practically. In fact, to take it a step further, this even assumes a probability measure at all, which would be tricky to derive from first principles. I think this is more your protest, and it’s certainly well-grounded as a concern.

    However, in order to make assertions in general, we don’t really need to know the exact probability measure so long as we know the measure-zero sets, which one can at least make some reasonable-sounding assumptions about. You don’t need a particular distribution for anthropic explanation of some sort of finite fine-tuning to come out if the universe were infinite. Of course, you do need a measure to effectively apply anthropics as a predictive tool more generally, which I think is where the greater issue comes in, but I think that’s a slightly different problem.

    Thirdly, of course, is the issue of whether the universe is infinitely large at all, which would be the point where curvature would come in, as well as some other assumptions (e.g., cosmological principle) which may or may not be justified…

  22. Peter Woit says:

    The discussion is about string theory as a TOE, and my claim is that, as such it is empty, predicting nothing.

    I just don’t see that “pick randomly, I can’t tell you why” has any more content than “I have no idea what’s going on here, pick randomly if you feel like it”.

  23. Neil says:

    “Of course, if an infinite multiverse does exist, some other Woit is out there saying the exact opposite.”

    Ah, so there is one invariant in the multiverse. All Woits are contrarians.

  24. Peter Woit says:

    I don’t think skepticism about whether the multiverse is science is contrarian, but rather pretty boringly mainstream and conventional…

  25. John says:

    Scott Church mentions that one possible reason for people supporting the multiverse is to defend against fine-tuning. Peter and others are clearly against the idea of the multiverse. So does that mean Peter is not afraid of fine-tuning, supports it or has another explanation?

  26. Peter Woit says:

    It bears repeating that I’m not “against the idea of the multiverse”. Maybe there is a multiverse, I have no idea. My claim is rather that current multiverse theories have no significant scientific content.

    As for “fine-tuning”, I just don’t think there’s a real problem there. I have no idea what is responsible for the CC, so no idea how to calculate it. People who think they have a unified TOE, in which the CC is calculable, may have a problem with it generically being determined by the Planck scale and thus way too large. They have a fine-tuning problem.

    In practice though, such people don’t actually have a viable TOE with a calculable CC, so what they do is say “my TOE has a huge landscape, so I can’t actually calculate the CC, instead I’m just going to assume it’s equally likely to be anything”. When they do that, if they assume things are calculable in terms of the Planck scale, they then have a fine-tuning problem (which they claim to solve with a multiverse/anthropic argument).

  27. Scott Church says:


    When I said that many folks embrace string landscape/eternal inflation (SL/EI) out of a fear of fine-tuning, I didn’t mean to imply the opposite–that those who are skeptical of it don’t consider fine-tuning to be an issue. My only point was that many people have latched onto it unquestioningly for reasons that are ultimately less than scientific. Few physicists dispute the seriousness of fine-tuning issues, but many think that attempts to account for them via SL/EI have been at best unproductive, and at worst disingenuous. Too many people today just default to them because they’re fluid enough to explain away what we see rather than explain it, and inappropriately blur the distinction between physics and metaphysics in the process. What’s needed are models that are specific and robust enough to make predictions we can verify, and our search for them must be philosophically as well as scientifically honest. In this Brave New Post-Empirical world we’re in grave danger of losing sight of that.


    Regarding multiverse probability measures, one of the best treatments of the technical issues I’ve seen was given by Ellis, Stoeger, and Kirchner (2008). But regardless of how we handle these issues we still have at least two fundamental problems. First, there are serious issues with an infinite universe, philosophical and observational, and even if there weren’t I don’t see how one could ever be verified (even in the case of negative curvature we’d only have infinite spacio-temporal extent if the universe/multiverse was simply connected topologically–an untestable premise). Second, the entire question of multiverse probabilities is moot until there’s a demonstrable physical method for actually producing one. To that end, the only game in town is eternal inflation, which brings us right back to the whole too-fluid-to-be-useful problem. It seems to me that until there’s a way around these conundrums SL/EI is more metaphysical belief than genuine science… turtles all the way down. There’s nothing wrong with that per se, but if that is one’s view then one should be honest about it, and many today aren’t. 🙂

  28. Hank Schraeder says:


    All this stuff is horseshit and I wish you’d stop reporting on this nonsense. You’re preaching to the choir, so find something else to write about. I wish these so-called researchers stopped doing this garbage. They’re trying too hard to justify their overly big paychecks. What nonsense.

  29. Layman 2000 says:

    @Hank Schraeder:

    You are wrong. Woit is not preaching to the choir. As a layman interested in physics I hear a lot about multiverse these days and even bought and read a book about it which turned out to be totally unsatisfactory. So I went out to look for arguments and counterarguments, and this site is an important source of information and informed opinion.

  30. Simeon Every says:

    The question of whether multiverse theories are scientific depends on what ‘science’ is taken to mean.
    I always assumed that science meant ‘that which can be known using scientific methodology’, but I sense that some subscribe to a canonical interpretation of science, whereby a theory that invokes entities consistent with established entities that have been accepted as scientific is by default a scientific theory.
    I have no good philosophical argument against the canonical interpretation but can see that it is of no obvious practical value.

  31. Peter Woit says:

    Simeon Every,
    I don’t think “invoking entities consistent with established entities” gets you into the “canonical interpretation of science”. For example, few agree that my favorite theory of pre-big bang physics (a big green turtle did it) is science even though it satisfies your consistency criterion.

    A more common “canonical interpretation of science” is that it’s whatever scientists do. Susskind in particular argues for this. It’s also why it’s important that scientists speak up and challenge those in their ranks who are on a campaign to get acceptance for pseudo-science. I agree there’s a good argument that there’s too much on this blog about this, but an equally good argument that someone needs to be doing this, and at the moment far too few physicists are stepping up.

  32. AcademicLurker says:

    “Of course, if an infinite multiverse does exist, some other Woit is out there saying the exact opposite.”

    This sounds like the motivation for a Woit based research program, in which physicists postulate different string vacua and then try to predict what Peter Woit would think of the multiverse in the resulting universe…

  33. Code Ferret says:

    @AcademicLurker: That gave me a hearty chuckle this morning. Thanks

  34. Anonyrat says:

    In no branch of the multiverse does multiverse theory have significant scientific content, and therefore in all branches, the Peter Woits there will be saying this. That is what is invariant.

  35. Magnema says:

    @Peter Woit: I agree with your issue; however, I think the same question you raise can be applied to classical statistical mechanics – in a purely deterministic system, from whence the statistics? As I recall, that’s a serious problem in the philosophy of physics as well – but I’d guess they’re one-and-the-same problem, namely, how do you get from ignorance (either of the microstate of a system or the self-locating ignorance in the multiverse) to probabilities? Yet we do like to think we have an idea of what’s going on in statistical mechanics.

    @Scott Church: For the first question (infinite universe), I admit, I have no response, although I would say that (as a Bayesian) the longer we go without observing topological complications, the higher we may want to raise the probability of a simple topology… but that depends on your priors over topologies, which depends again on a measure over possible universes. I’m also curious about your “problems with an infinite universe” – aside from being untestable and the open question of how to derive probabilities in the infinite case (which I’m not sure is a “real problem” in the same sense as what I said to Peter – it’s no different than the usual foundational Stat Mech difficulties).

    For the second, I would have two responses. Firstly, “argument without a particular mechanism” can be valid sometimes – after all, it is the entire principle behind (ordinary) inflation. If it works there, one shouldn’t necessarily reject the multiverse idea on that ground (although I admit that it weakens the hypothesis, just as it does inflation). Secondly, as I said before, I think anthropic argument may work even when you don’t understand the probability measure well enough to make accurate numerical predictions, although the latter would be a key element that would turn you from “a plausible-sounding metaphysical idea, which resolves a particular problem” into “a proper physical theory.”

  36. Simeon Every says:

    The most contemporary developments in philosophy of science (that I am aware of) characterise research paradigms in terms of the extent to which theories in given paradigms produce testable predictions. This seems to be favouring methodological issues over canonical issues. Unfortunately there seems to be a prevailing mood amongst theoretical physicists (at least among those with the most public prominence) to declare philosophy irrelevant.

  37. anon says:

    Magnema: “…how do you get from ignorance (either of the microstate of a system or the self-locating ignorance in the multiverse) to probabilities? Yet we do like to think we have an idea of what’s going on in statistical mechanics.”

    I submit that we have less of an idea of what is going on in Statistical Mechanics than we sometimes pretend to. Regarding the “Everything that is Allowed is Mandatory” problem, much of classical Statistical Mechanics take some form of ergodic hypothesis as a postulate, but this is mostly unproved and is often technically false (ergodicity often only holds only modulo a coarse-graining assumption). There are some rather interesting and complicated (even chaotic) classical systems that are not ergodic. See for example the Fermi–Pasta–Ulam problem. These may be the rule rather than the exception.

  38. Low Math, Meekly Interacting says:

    Simeon: I don’t have much of an opinion on the relevance of philosophers, except to say a robust research program doesn’t need them. I would furthermore posit if your research program could stand to benefit from an infusion of philosophical rigor, that’s a sign of deep trouble. History ought to be an adequate guide, and it tells us science disconnected from observation is a dead-end path. There are precisely zero examples of true success in the absence of empirical evidence. There is precisely zero reason to think true success can be found in the absence of empirical evidence in the future. If quantum gravity and/or unification research cannot make contact with experiment, then they have no future as science, or should not, in my opinion. That’s why the anthropic multiverse is anathema. It’s an attempt at creating a loophole for a line of inquiry that may very well be hopeless. It’s an excuse to keep working on it.

    To my silly little mind, it really looks about that simple. I’m sure a philosopher could frame the position much better, but is there much added value? It’s a question of whether to stop and focus on things one can ever test, or redefine “science” for oneself to avoid that requirement. What more is there to it, in actuality? Very little of true substance, it would appear.

  39. Scott Church says:

    Hi Magnema,

    Beyond the untestability issues, the serious problems I was referring to were; 1) The philosophical problems with instantiating actual (as opposed to conceptual) infinite universes in the real world (e.g. “Hilbert’s hotel”, etc.); and 2) Physical difficulties with the formation of spacio-temporally infinite child universes in eternal inflationary scenarios (for instance, see Ellis & Stoeger (2009)). The former in particular is a hotly debated issue and both are off-topic here. My intention was simply to note them in passing.

    As for inflation, I’m with you in that I also think its current fluid state, by itself, shouldn’t lead us to reject Level I multiverses outright. I must admit, I have something of a love/hate relationship with inflation. On the one hand, it’s a beautiful paradigm that solves multiple seemingly disparate fine-tuning issues in one fell swoop (e.g. flatness, CMB isotropy, etc.), and it is testable in the sense that specific characteristics of the CMB and model-specific predictions from tunneling and/or reheating scenarios allow us to rule out many inflationary models. It also seems to be the only viable option for bringing at least some multiverse scenarios into the realm of observational testability… in principle, that is. However, in practice it only pushes the fine-tuning problems back a level. And in the absence of any clear inflaton candidates it leaves us free to dream up any scalar field scenario we like (hence the fluidity). Yes, a conceptually valid physical mechanism is involved and those models can be tested, but the end result is little more than whack-a-mole. And to many folks the real appeal of Level I multiverses is that they open the door to possible anthropic solutions to fine-tuning problems. But I don’t see how they can do this without the string landscape–all of which is true whether we’re able to define valid probability measures on the outcome space or not…

    Which sadly, brings us right back to square one… untestability. I must say, I was one of those who eagerly awaited the Planck results hoping for the “smoking gun” of primordial B-mode CMB polarizations. I was also one of those rejoicing when the BICEP2 results were first announced. But once those results started to unravel, it was disconcerting as hell to watch Linde and others furiously backpedal on B-modes as a smoking gun. “Geez,” I remember thinking. “Steinhardt is right… they really can get whatever they damn well please out of this framework, can’t they…?” I’m no expert on these matters, but I suspect that we aren’t going to find a viable inflaton candidate or be able to meaningfully evaluate string theories without probing much closer to Planck scale energies. I don’t see how this can ever be done without an LHC the diameter of the solar system (that has all the attendant luminosity problems solved as well). Even if it were possible to build and operate such a device, imagine what it would take to get anyone to fund it… especially if Republicans control Congress. (Sorry… I couldn’t resist. 🙂 )

    It may not sound like it so far, but my real complaint isn’t with inflation, multiverses, or even string/M-theory per se. It’s with all the high priests of “post-empiricism” who expect us to flippantly dismiss all these concerns and line up behind them so they can redefine science as a beautiful, Utopian fairy tale land… with verdant landscapes and lush theoretical gumdrop trees… where gossamer winged math fairies grant every wish… where 72-virgin review committees publish every paper and dry every tear… and nothing as pedestrian as real-world facts are allowed to spoil anyone’s party. Unfortunately, the trouble with fairy tales is that sooner or later one wakes up from them… and more often than not, with one helluva hangover and someone pressing charges. 😀


  40. Carlos Looney says:

    Neil: whether there are other universes out there in a multiverse with other “Peter Woits” saying different things depends upon your definition of a “Peter Woit.” Do we mean just the DNA structure, what about environmentally induced alterations, mental content, or other variables?

    According to some Physicists of the MWI persuasion, each “Peter Woit” is constantly splitting into two Peter Woits in two different worlds everytime a measurement is performed. Should we distinguish between these Peter Woits and what criteria? Are the “Peter Woits” we do not encounter real or unreal? Notice that the Peter Woits in the inflation scenario are real Peter Woits. But with MWI it is an open and probably never resolvable question.

  41. Peter Woit says:

    Actually I think the multiverse is an opportunity for philosophers of science to show that their subject is relevant and useful. There’s a lot of material there to apply their best thinking about the demarcation problem to. Unfortunately I haven’t seen much of this happening.

  42. Low Math, Meekly Interacting says:

    Peter: I’m probably too dismissive or pat in my thinking, but sometimes professional philosophizing seems like a very convoluted path to the obvious point. Book-length treatises which amount to the substance of clever aphorisms like “beautiful horses win races.”

    But you’re probably right about programs that can keep moving the goalposts of testability due to an abundance of flexibility. It’s one thing when a model’s predictions are beyond the grasp of mere mortals to test. But what about those that could be, but also could be testable at, say, any energy between what the LHC can probe and the Plank scale? One can call it ugly, but if beautiful horses win races, proponents seem initially to be game for the challenge of science, and I can think of no iron-clad argument for denying their legitimacy at the outset. Of course, if the race is allowed to go on forever, that’s a different story.

    I.e., there are research programs that start out being arguably scientific, but evolve through repeated failures into something else. My challenge to the philosophers would be to give us a clearer answer to the demarcation problem, one that can tell us the right time to say enough’s enough, and have everyone agree.

  43. Magnema says:

    @anon: Agreed on the lack of understanding of Stat Mech – the more I learn, the less I feel like I know about how it emerges. My point was more to make the analogy between established science (with, admittedly, poorly understood foundations) and multiverse-type arguments. I suppose I am assuming something analogous to the ergodicity in the multiverse case as well.

    @Scott Church: I agree with everything you just said.

    However, I note an issue here on Bayesian vs. Popperian science. In the former, which is usually regarded as the stronger theory, the CMB tests were taken as evidence, rather than a total disproof. It would seem that those individuals place a very high prior probability of inflation/string theory/related hypotheses, which means that, while the test should have been sufficient to bump an “uncertain” person into a “very certain” position, the converse is insufficient to bump a “very certain” position into anything less than a “moderately less certain” position. (E.g., suppose that they think: 100% chance of positive results meaning inflation, 99% negative results do; then, while they should reduce their confidence in their theory from, say, 99.999%, it’s not going to be reduced too dramatically.) Now, whether or not their utter confidence in their theory is justified (I tend to think not) is a different concern, but it can still be scientific seeing as the prior probabilities have no scientific way (to my knowledge) of being determined…

  44. Jason says:

    Hey Peter,

    I’m a former physicist; I got my PhD working on CMS before the LHC turned on, finishing in 2006. Although my research was focused on rare signature events, specifically Z’ searches for indications of Super Symmetry, most of my colleagues and I thought Super Symmetry was probably made up non-sense. We all knew the reasons the theorists wanted it, but from an experimental perspective, it seemed a big stretch that there would be twice the number of particles, and not a single bit of evidence that they impact any Standard Model (SM) measurement. Wouldn’t it be difficult to exclude the idea that those extra SM particles would have changed the result of any SM measurement going on already? There are many high precision measurements of the SM, surely something would have shown up by now, not just in the high-energy experiments?

    I have been enjoying your blog for a while now. I don’t really have a strong opinion one way or another about the multiverse. However, I do think your arguments are a bit more persuasive for me that what I’ve heard on the other side. I’m just wondering, if it were for the practice of debate, what would be your best defense of the multiverse? Or in other words, what would be your best arguments against yourself?

    Thanks Again!

  45. adrian says:

    Dear Peter,
    I have followed the discussion about the topic of the landscape in this blog post. Before, I asked you to qualify a statement you were quoted on—see message above, you did it. I probably disagree about the total lack of value of string theory as a theory of Everything, but the topic in itself [I mean TOE] interests me very little.

    I wanted to comment something, probably a bit influenced by my daily read of the hep-th arXiv. As you may guess, I do work on string theory, publish regularly and never worked on ‘the landscape’. I may be a bit unfair in the comment that follows.

    It is my impression that most physicists or most string theorists are actually NOT working on landscape related topics. So, when people refer to the ‘multi-verse mania’, I find hard to understand what they are talking about. It seems to be something that most goes on the press than on the actual work of physicists/string theorists.

    Let me take as an example the work of two physicists I know personally and have discussed at length with them: Matt Kleban and Ben Freivogel. They are both solid and serious physicists. It is a pleasure to discuss any particular problem with them, they will always make an interesting observation–probably also a correct one– and one always leaves enriched after a conversation about Physics with them.

    Having written this, we open the spires page for their work. I will take them as representatives of physicists working on landscape related topics. Again, may be there are other representatives, but these two, at least, I know for a fact are serious and will not lie in a paper. As you can see, their work on the topic is not excessively cited by colleagues. It is just cited like any other set of papers of other colleagues. I mean, papers cited order 60 times since 2008 is not reflecting a real rush of activity, many of my more recent papers have more citations than that. Notice, this is NOT an evaluation of the quality of the papers of these two physicists. I -know-the papers are very good. This is just a reflection that there is not a lot of activity. It is not the case that large amounts of money in funding are deviated into that topic.

    It is certainly NOT my intention in this comment to offend anyone, much less Kleban or Freivogel. I am just taking them as representative of a line of work [actually as serious physicists working on that line]. I do know that a very cited paper does not mean it is a good paper, but it reflects the level of activity on the topic, sometimes, it also reveals the paper is very good, but is not always the case. I know many papers that are very good and almost not cited.

    Could you please let me know if my view of this topic is misguided? Let us also observe that of Susskind or Linde say ‘multiverse’ many times, this does not mean the topic has wide acceptance among the majority of working physicists. Does not mean either that big amounts of funding are assigned to that area. I may be wrong, could you please correct me? I do not have good data.

    Having written this, I believe the criticism you voice on the topic is healthy.
    I have the impression that you sometimes get carried away and involve the whole area of String Theory under the same criticism. This is pernicious. But it is your blog and are your opinions.


  46. Peter Woit says:

    I think your argument about why to not expect SUSY at the LHC (zero evidence, direct or indirect, for a single SUSY state, out of a large expected number) is one a lot of physicists shared. It’s too bad that it rarely got made publicly.

    I’m not much of a debater… The best argument the multiverse people have is the one for the CC. As for the landscape, you can try and argue that research should continue, absent a successful idea. All in all though, I think arguments for the multiverse are even weaker than arguments for string theory.

  47. Peter Woit says:

    You’re right that most string theorists don’t work on the multiverse, and pretty much do their best to try and ignore it. Many of them do however invoke it as an excuse when challenged about the failure of string theory TOE ideas to work out. Many of those promoting the multiverse are not string theorists (for example Carroll and Tegmark), but physicists who want to address “big ideas”, and end up falling for something empty which seems to do this.

    It is a fact though that the string theory landscape and the multiverse get a huge amount of public attention. Good string theorists may not take them seriously, but they’re also not willing to go on record saying anything critical about the subject (there’s a strong ethos of “string theory is under attack, so thou shall not criticize anything about string theory or the arguments being used to justify it”). By letting this go on and doing nothing, they are allowing some serious damage to be done to their field. There are a lot of people out there that the press and the public rightly turn to, and they’re not stepping up to the job (put differently, why aren’t there more like David Gross?). I confess to a certain level of annoyance about this, that I’m doing something unpleasant because the people whose job it is to deal with it don’t want to.

  48. adrian says:

    Dear Peter,
    I would like to briefly comment on your answer above. You stated
    ‘Good string theorists may not take them seriously, but they’re also not willing to go on record saying anything critical about the subject (there’s a strong ethos of “string theory is under attack, so thou shall not criticize anything about string theory or the arguments being used to justify it”) ‘

    I have two observations:

    1–that many string theorist do not go ‘on record’ with their opinion about the landscape and anthropic arguments because they have not worked on the area actively. It is difficult to have an opinion and record it somewhere if one have not contributed or deeply thought about the thing. Do you agree? In my case, I see anthropic arguments with some fascination-rejection (for example, I find the book by Barrow and Tipler beautiful to read). I see the landscape as a VERY interesting geometrical problem. But since I wrote no paper on any of those areas, all I can contribute are ‘feelings’. Why should I go on record or on the press to say my ‘feelings’?

    2–The area of String Theory is, indeed and unfortunately, under attack. The attack that people who are NOT working in the area are expressing is sometimes short sighted, sometimes ill-intentioned and sometimes based on misconceptions. Sometimes it is fair criticism. This is a type of ‘high-profile’ attack that other areas of Physics [QCD, Collider Physics, Condensed Matter] do not suffer from–I mean press articles, books, blogs. Believe me that the most interesting criticism of the topic, I have heard from physicists working in the area. Indeed, I find that the fairest criticism is made by those who are actively working on the topic, in seminars and papers. Of course, some other people write press-articles ”in favour” of String Theory that are pathetic to read.
    But as I wrote some months ago in your blog’s comments—that time about Amanda Peet’s talk at Perimeter: the science is made in papers, seminars and conferences, not in blogs, not in press articles nor in popular talks.

    thanks for the attention

  49. Lucy M says:

    Adrian “It is difficult to have an opinion and record it somewhere if one have not contributed or deeply thought about the thing. Do you agree?”

    Yes…it’s a fact of life that the people immersed in a theory understand by far the most. Would you agree that this is the same for every theory that has ever existed?

    A serious problem of this, is that while people on the inside of a theory may be totally impartial, sometimes they are not. This is a serious problem that bedevilled knowledge before and outside the scientific revolution. Knowledge inevitably gave power to those who preached it.
    Prediction is there among other reasons to prevent scientific incumbent knowledge from having to adopt new theories on the word of its insiders, and they from becoming philosopher-kings in the event their theory is adopted.

    Can I just say that it is very clear that String theorists have had ‘group shared experiences’ that leave them with a shared sense of sureness Strings are the right way forward.
    The problem with group experiences like this, is that the sense of sureness has no skeptic on the inside, at all. This means there is nothing to check excesses. The attempt to re-define Science without predictions, and with a set of traits exactly matching what string theorists think they can demonstrate, is an astonishing example of what can happen.
    But also the quality of argument from string theorists….without internal criticism for standards, these arguments have become really superficial and liced with fallacious debating devices. Your most senior theorists – founders even – have come out with the most specious and frankly dishonest arguments.
    I believe this is a direct consequence of a group of people succumbing to group shared experiences instead of treating such things as par for the course. Because they are. Everyone has them in the realm of their own theory.

  50. Lucy M says:

    Adrian “It is my impression that most physicists or most string theorists are actually NOT working on landscape related topics.”

    Yes but the REASON is that there literally are zero research directions, the multiverse is perfectly sterile.
    What it doesn’t mean is that most string theorists BELIEVE or SUPPORT the multiverse.
    If they won’t come out and make a firm break with their colleagues that do support the multiverse, then – again – we’re being asked to trust the word of insiders bonded to each other by nebulas group shared ‘experiences’.
    Why aren’t you giving your full details? Can you post a link to a website set up by dissenting string theory insiders?
    I notice you use the subtly modified general vocabularly that is the hallmark of the ‘shared group experiences’. For example the word ‘beautiful’ gets used rather a lot. Another hallmark is more than a whiff of ego involving ideation not unlike those who have had religious experience.
    Another hallmark is avoidance of fundamental criticism, and of the tactics of own side.
    Apart from that you seem a very sweet person :O)

Comments are closed.