Falsifiability and physics

Symmetry magazine today published an article on Falsifiability and physics, yet another in the genre of defense of current HEP theory against its critics. As usual, only defenders of the status quo are quoted, the critics remain unnamed and their actual arguments ignored. I don’t completely understand this journalism thing, but if you are writing about a controversy, aren’t you supposed to contact people on both sides?

The problems with this article begin with the misleading subtitle: “Can a theory that isn’t completely testable still be useful to physics?” The problem here is not theories that aren’t “completely testable”, but theories that aren’t testable at all, that make no testable predictions at all.

The article starts out by discussing Popper and the supposed “falsifiability” criterion for what is and isn’t science, leading up to:

But where does this falsifiability requirement leave certain areas of theoretical physics? String theory, for example, involves physics on extremely small length scales unreachable by any foreseeable experiment. Cosmic inflation, a theory that explains much about the properties of the observable universe, may itself be untestable through direct observations. Some critics believe these theories are unfalsifiable and, for that reason, are of dubious scientific value.

Who are these “some critics”? Where do they say that the reason there is a problem with string theory is “unfalsifiability”? For the case of one critic I’m pretty familiar with, chapter 14 of his book is all about how “falsifiability” is not something that can be used to decide what is science and what isn’t.

We’re then told that:

At the same time, many physicists align with philosophers of science who identified flaws in Popper’s model, saying falsification is most useful in identifying blatant pseudoscience (the flat-Earth hypothesis, again) but relatively unimportant for judging theories growing out of established paradigms in science.

Unclear who “many physicists” are, who the “philosophers of science” are, and what flaw in Popper is being referred to.

In an odd move, the article then turns to the topic of SUSY, where the problem isn’t that well-advertised SUSY models (with electroweak scale SUSY breaking solving the “naturalness” problem) aren’t falsifiable, it’s that the LHC has falsified them. As usual in science, if your model gets falsified, instead of giving up and doing something else you can change your model to something less desirable that hasn’t been falsified (SUSY models with symmetry broken at higher energy scales) and keep on going. This is though what philosophers of science call a “degenerating research program”, which is not a good thing.

There’s more in the rest of the article, but actual critics remain invisible and their actual arguments unaddressed.

Update: Will Kinney has some appropriate comments.

Update: Massimo Pigliucci has posted here his contribution to the “Why Trust a Theory?” volume, which discusses “falsifiability” and the “String Wars”.

This entry was posted in Uncategorized. Bookmark the permalink.

33 Responses to Falsifiability and physics

  1. David Appell says:

    I wouldn’t call the writers for Symmetry “journalists.” They’re not and that’s not their goal or purpose. Symmetry magazine, by design, has an agenda. Moreover, it’s not reader supported:

    “Symmetry is a joint publication of Fermi National Accelerator Laboratory and SLAC National Accelerator Laboratory. Symmetry receives funding through the US Department of Energy.”

  2. GoletaBeach says:

    I don’t really see that article via the HEP theory lens. It makes some point or another about how for atoms falsifiability was an inappropriate criterion. I think the accurate point is: atoms were found because experimentalists kept at it and found them. Echoing an old essay by Luis Alvarez…. likely if there had been a review panel of theorists, they’d have never funded the experimental work that led to atoms being discovered, because of some philosophy or theory or prejudice… just like RT Cox’s discovery of parity violation in the 1930’s was not appreciated.

    From an experimental perspective, more beam energy has traditionally been the most effective way to make new discoveries. There was not a clear mass target until about the PEP and PETRA eras, where they hoped to make top quarks (and failed).

    Somehow the requirement of a vetted mass target seeped in through the review committee process now to all accelerator proposals.

    Best to just ignore SUSY or whatever and build the next machine. There will be (and always have been) naysayers who say it isn’t worth it. Well, the marketing budget in the US for tobacco is… almost $10 billion/year. Societally, a 100 TeV machine isn’t expensive. If nothing is discovered, of course, it would have been better to have spent the funds on air conditioners for elderly and impoverished people in US cities… or simple water purity in most of the world. If a shocking discovery that rewrites the foundations of physics is made… well… easily worth more than tobacco marketing.

  3. Peter Woit says:

    The distinction I think you’re trying to make between journalism and PR is not so clear here. I’ve dealt a little bit with people at Symmetry, for instance concerning this article
    It’s not so easy to distinguish what they do from what Quanta does (other than that I bet Quanta has a lot more money…), although perhaps the Simons Foundation tries to keep Quanta’s coverage separate from promoting their own research, while Symmetry may try to help promote DOE-funded research.

    In any case, even if Symmetry were strongly agenda-driven in support of Fermilab and SLAC scientists, I don’t think it would be hard to find such scientists with critical views of string theory and SUSY. So the question remains, why just report one side of a controversy?

  4. Peter Woit says:


    I more or less agree with you: ignore the theorists and, if at all possible, build a higher energy machine. This article isn’t though at all about that, no experimentalists appear. There are lots of topics in the article, some of which have nothing to do with any controversy over falsifiability (e.g. I don’t think dark matter models and testing them is particularly controversial).

    Leaving inflation aside, the controversial topics are string theory unification and SUSY, and the fundamental controversy is over how you evaluate the results of this research. The results now look like a failure if evaluated by any conventional scientific standard. The claim that falsifiability is an inappropriate way to evaluate these ideas is not then supplemented by any suggestion of how they should be evaluated. A lot of this article looks like an attempt to basically argue that these ideas should be immune from evaluation.

  5. Amitabh Lath says:

    The article points out that for Popper unfalsifiable theories were Freudian psychology and Stalinist history. Surely you don’t think SUSY belongs with them? It is falsifiable, you may need equipment and techniques that are beyond our capabilities (for now) but that’s not the theory’s fault. Neutrinos were considered unfalsifiable by Fermi.
    As for string theory, AdSCFT etc. doesn’t pretend to be about our reality so the where do you even start with the falsifiable argument? You could argue it’s been N decades and they really should have gotten the math straight by now and made some testable predictions but they could counter with “there aren’t enough people working on it, we need more string theorists.”

  6. martibal says:

    It seems hard to counter with “there is not enough string theorists”. By a very rough approximation, I would say that the number of string theorists is one order of magnitude bigger than the number of people working in the second best known approach to quantum gravity (loop quantum gravity), which is again one order of magnitude bigger than any other approach to quantum gravity/unification (dynamical triangulation, noncommutative geometry etc).

    Do the quality of the results obtained in these various theories range in the same order ? Not sure that the number of people working on a theory is such an accurate criteria.

  7. Davide Castelvecchi says:


    I enjoy reading Symmetry and I have written for Symmetry earlier in my career. People who work there have standards of integrity. But they are very conscious of the fact that what they are doing is not journalism. It’s science communication, and often of very high quality, but it’s not journalism. Quanta is a whole other story. It is philantropist-sponsored — which comes with its own caveats — and tends to take the side of certain communities, but it still explocitly strives to uphold journalistic standards.

  8. Peter Shor says:

    @Amitabh Lath:

    If, as you say, “string theory and AdS-CFT doesn’t pretend to be about our reality,” why are so many physicists convinced that AdS-CFT shows that quantum gravity in our universe has to be unitary?

    And there aren’t many physicists in Physics departments working on different theories of fantasy physics, while there are tons of them working on string theory and AdS-CFT. Shouldn’t fantasy physics be studied in Philosophy, Mathematics, or Creative Writing departments, rather than Physics departments?

    This is just a fairly transparent ex post facto excuse for why string theory and AdS-CFT have failed to say anything useful about physics in our universe.

  9. Dom says:

    This paragraph quoted below has me puzzled because it implies that somehow falsification is impeding people working on new ideas they know are likely to be wrong (which tends to take care of the falsification bit it seems to me).
    “Tracy Slatyer of MIT agrees, and argues that stringently worrying about falsification can prevent new ideas from germinating, stifling creativity. “In theoretical physics, the vast majority of all the ideas you ever work on are going to be wrong,” she says. “They may be interesting ideas, they may be beautiful ideas, they may be gorgeous structures that are simply not realized in our universe.””

  10. vmarko says:

    From the article:

    “One such theory is cosmic inflation, which (among other things) explains why we don’t see isolated magnetic monopoles […]”

    All the falsifiability stuff aside, I still find it hard to believe that serious people consider this as a valid argument in favor of inflation. In addition to magnetic monopoles, the cosmic inflation could very well be able to explain why there are no unicorns in nature. Would that increase anyone’s confidence in the theory?

    The lack of magnetic monopoles is “explained” (basically by definition) by Maxwell’s classical electrodynamics. A serious scientific/skeptical attitude should be that — absent any compelling arguments that something ought to exist — the default position is that it doesn’t exist. The burden of proof is squarely on the person who claims that something ought to exist, not the other way around.

    Of course, the magnetic monopole waters were muddied by the story that GUT’s give a compelling argument that monopoles ought to exist, so their absence becomes something that should be explained. However, by far and large, GUT’s have been falsified by lack of proton decays, which (among other things) disqualifies their argument regarding the existence of magnetic monopoles. You shouldn’t seek to explain something that is predicted by a theory that is known to be wrong.

    Are there still that many die-hard GUT-supporters out there, who keep the magnetic monopole argument alive, despite all proton-decay experimental results? Or is it just the case of people parrott-repeating arguments from books which are half-a-century old and haven’t been updated with more recent data? Or am I missing something obvious here?

    Best, 🙂

  11. Low Math, Meekly Interacting says:

    I find the whole notion of this debate distressing and always have. I know of very few people who are fundamentalist in their belief in the “scientific method”, but for me some hope of contact with experiment is a non-negotiable. I simply can’t fathom the notion of “science” without it.

    Of course there needs to be toy models, simplifications, abstractions and so forth to attain the goal of observable consequences. But if the goal of modeling nature as it is, in a both descriptive and predictive manner, is dispensed with indefinitely in pursuit of other criteria for success which are necessarily human constructs, then in my opinion hope is lost. Human judgment simply cannot be trusted indefinitely, and humans should not feel offended by this truth. It is in our nature to err, and we cannot avoid it without external guidance. If the universe exists without us, then it is our only ultimately reliable guide, and if we cannot extract the information we need from it, then I think it is ultimately folly to try.

    That might mean giving up on answering some questions. The fact that many find this unacceptable may be a major source of the current predicament.

  12. Peter Shor says:


    I think that one of the ways physics has advanced over the years is that people come up with all sorts of crazy ideas (relativity, quantum mechanics, Dirac delta functions, the replica method, Feynman diagrams, all certainly seemed crazy to some people when they were first proposed), and only kept the ones that agreed with experiment. Without experiment, physicists have no way to discard the incorrect crazy ideas and keep the correct ones.

    Tracy Slayter objects to using the falsifiable criterion, as it will keep physicists from coming up with crazy ideas in the first place, but I don’t think that’s the real problem.

    I am beginning to think that the problem with HEP theory currently is not that physicists don’t discard non-falsifiable ideas, but that without experimental input, they don’t have any reasonable way to vet their crazy ideas, so they end up throwing out the baby and keeping the bathwater. There is certainly some sociological process going on that selects some of the crazy ideas, and reject others.

  13. Peter Woit says:

    Amitabh Lath,
    As I keep trying to explain, I’ve never been someone who thinks “falsifiability” is the issue here and in particular re SUSY it’s a sterile debate. Amidst SUSY models you can find whatever you want: falsified already, falsifiable now, falsifiable at HL-LHC, falsifiable at FCC-hh, falsifiable a hundred years from now, unfalsifiable. The only clear relevance of the falsifiability criterion here is that people should stop paying attention to the small number of Gordon Kanes of this world who, in the face of repeated experimental falsification of an idea of theirs, refuse to give up on it.

    Re SUSY, I think the best advice now for experimentalists is to ignore most of what theorists have to say about the virtues of this or that SUSY model, and just ask the pragmatic question: does the model provide a useful target for designing searches that will be sensitive to a significant range of possible new physics?

  14. Peter Woit says:

    Agreed about the weirdness of “inflation explains no GUT monopoles” argument, given that the relevant GUT models have failed other tests, so the simplest explanation for no GUT monopoles is that the GUT models are wrong.

    One sorry aspect of these debates is that arguments made long ago for string theory/SUSY/GUTs/etc. do keep getting repeated, long past the time when argument has failed. A lot of this is people just repeating things they read long ago, unaware of what has happened since. Some of it is people who should know better, but for whatever reason can’t let go of a falsified argument.

  15. DB says:

    this might be of interest to you.
    Short video. The magnetic monopoles stuff starts at 02:30.


    Peter (and the rest):
    Why wouldn´t the top theoretical physicists let go of some theories which, according to themselves, do not describe reality?
    It´s certainly not because of money (they´re all sorted…).
    Can anyone think of another reason?

    Finally, this 2014 interview with Susskind is very interesting.
    From the 50th to the 55th minute he talks about No Symmetry, No Beauty, and …
    letting an idea/theory go when it´s not correct!!


  16. Peter Woit says:

    Dom/Peter Shor/LMMI,

    I don’t think anyone is seriously arguing that theorists should only work on ideas falsifiable now or in the near future, and I also don’t think those working on highly speculative ideas often argue that it doesn’t matter whether what they are doing ultimately leads to something testable. The real problem is not “falsifiablity”, but how you evaluate the progress of speculative research programs. You can’t do this by demanding an experimental test, but you also can’t just accept theorists’ assurances that people should keep supporting them to continue work on their favorite idea of thirty years ago, despite all evidence it hasn’t worked out.

    What’s disturbing to me is that, increasingly, the string unification/SUSY research program seems to have moved from “evaluate us by LHC results or progress on these crucial problems that are in between us and a testable theory” to “there is no way to evaluate us, you just have to believe us, because there are so many of us and we’re so smart.” That’s not the way science is supposed to work, for good reason.

  17. Peter Woit says:

    What I’ve learned over the years is that, once people become devoted to a certain idea and invest a lot of their time, energy and public reputation in it, they are highly unlikely to ever abandon it, no matter what new information comes along and no matter how smart they are. The best you can expect is “I still believe there must be something right about that idea, but I’m now spending my time doing something else….”

  18. DB says:

    thanks for the reply.
    Your last sentence is key for me.
    I´m starting to think that many of the top string theorists have sort of “given up”, and are really working on something else.
    Though they don´t want to come out and say it in public, for obvious reasons.
    Which, honestly, leaves them in a bit of a dishonest situation.

    It will be very interesting to see how many of the Strings 2019 conferences in Brussels next July will focus on ST itself.

  19. Anonyrat says:

    IMO, if interesting mathematics is emerging from a speculative line of investigation in theoretical physics, and if “interesting” mathematics has some reasonably objective meaning, then that might constitute a measure of progress sufficient for the decision of whether to fund those theorists.

  20. Peter Woit says:

    I’m all in favor of supporting theoretical physics work that leads to interesting new mathematics. Keep in mind though that evaluating what is interesting new mathematics is something you need mathematicians for. Most physicists are likely to take the attitude that work that needs to be evaluated by mathematicians should be funded by the mathematics part of NSF. They may even take the attitude that theorists doing this kind of work should be in a math department. There are a few places (e.g. the Simons Center at Stony Brook) where there’s a healthy overlap of math and physics research, but unfortunately that’s not the case most places.

  21. vmarko says:

    Anonyrat and Peter,

    There are some countries (like Portugal) in which the academic system is set up so that any non-experimental physics (i.e. all theoretical physics, not just hep-th) is a part of the math departments, rather than physics departments. IOW, if you want to be in a physics department, you need to be working in a *lab*, and measure something with some equipment. Everything else, any pen-and-paper work, is considered math.

    The results of such an arrangement are, mildly put, catastrophic. The applications for projects, results that are obtained, papers that are published, etc., are all evaluated by committees in which theoretical physicists are present, but mathematicians are a majority. A paper published in a journal with “Math.” in its name is more worth than a PRL. And if you ask a mathematician to judge the work done by a theoretical physicist, they will mostly laugh at it. The levels of rigor, precision, and definition-theorem-proof style of writing, that mathematicians are used to, are something a theoretical physicist can never even hope to reach. Mathematicians by and large regard theoretical physicists as amateur wannabe-mathematicians, who are toying with some equations and delude themselves that they are doing something serious. They often dismiss the work of physicists just because it is not formulated as a theorem and proved rigorously, even without looking at what was actually done. And they often have a hard time understanding why the work of a hep-th physicist cannot be formulated as a theorem… I was often asked stuff like “How do you even know what you are doing, if you cannot explicitly spell out all your assumptions for me, before you make a statement what the result is?”

    Theoretical physics is not math. The whole math mentality is completely different from physics mentality, with different goals, intuition, background education, ways of thinking, and levels of rigor. There is just no way to consider theoretical physics as part of proper math, and it’s a Bad Idea ™ to even try.

    Best, 🙂

  22. Bernhard says:


    Also, one point you have mentioned ad nauseam that I have never seen been addressed in this or any article is the question of the negative progress rate of String Theory as opposed to whether is falsifiable or not. String Theory could become a falsifiable tomorrow. But it’s pretty clear it won’t be. Not tomorrow, not ever. Every time I read those types of articles people discuss with a straw man never addressing the real issue.

  23. Amitabh Lath says:

    Peter Voit, Peter Schor, the longevity of string theory is not due to the middle-aged practitioners you mention but kids in their early 20s who continue to choose to go into the field. Some of the best undergraduate students in our high energy experiment group have over the years chosen to go to grad school in theoretical physics 🙁

    Some go into phenomenology but some are indeed doing string theory.

    These students are the most smartest and most sensible I have ever met, the cream of the Garden State. They devour the literature, they are fully aware of the arguments on all sides. I cannot in any seriousness entertain the idea that they are led astray by hyperbole. I believe all the arguments about string theory not having made any progress in decades, not producing any testable results, being stuck in a made-up universe nothing like our own reality; these are not deterrents but attractions to this type of student.

  24. Peter Woit says:

    Amitabh Lath,
    The intended audience for the typical promotional string theory hype-fest piece is not ambitious youngsters considering a career in particle theory, but physicists in other fields, people in other sciences, university administrators, interested laymen, etc. In general, people who might have something to say about whether string theory research gets supported, and aren’t in a good position to evaluate the claims being made themselves. When you’re reading an article supposedly about some controversy, but only one side is quoted, what you’re reading has an agenda behind it.

    The topic of what’s going on with talented undergrads who are interested in theoretical physics, want to study it more deeply and possibly go to grad school and try to make a career in the subject is an interesting one, but not so relevant to articles like this. Yes, they aren’t going to impress a smart student seriously studying particle theory, and such students now do hear both sides of the controversy over string theory.

    I know quite a few such students, some pretty well, either through teaching them here at Columbia and then following their later progress, or through having students from elsewhere contact me for one reason or another. What I see happening now (at least in the US) is that the best students are, as always, going to a small number of the top graduate programs (e.g. Harvard, Princeton, Stanford), where most of the theory faculty often identify tribally as “string theorists”, but are now working on topics in GR/QFT/quantum information, etc. that have nothing to do with quantized strings or with string-theory based unification. The odd thing I keep hearing is that such students arriving at such a grad program are encouraged to spend a lot of time studying actual string theory (e.g. by reading Polchinski’s two volumes) to prepare to start research, even though the research likely won’t use any of this. My impression is that a lot of the theory faculty are unsure themselves which way the field should be going, and by default are suggesting students start off the same way they did 20 years ago.

    Studying theoretical fundamental physics attracts, for good reason, and despite the field’s problems, some of the best students around. Sometimes I do fear though that one reason the field is not generating much in the way of new ideas is not so much that the best students are drinking string theory kool-aid, but that they’re getting their entry into the subject of fundamental theory in a way that hinders rather than helps them come to grips with the real problems of the subject.

  25. Peter Shor says:

    Amitabh Lath:

    When it looks like two recent papers in a field are incompatible (despite both of them being quite reasonable papers on their own), and talking to some people in the field, nobody even seems to have noticed this fact, I suspect that something has gone seriously wrong with the field.

    The two papers are (a) the ones suggesting the the CFT in the AdS-CFT correspondence is a quantum error correcting code and (b) Shenkar and Stanford, “Black holes and the butterfly effect,” which starts out essentially saying “make a small perturbation to the CFT”.

    If you make a small perturbation (or even a large perturbation) to a state in a quantum error-correcting code, for any kind of quantum error-correcting code I am familiar with, you are extremely likely to be taken out of the quantum error-correcting code, which means that — as far as I can tell — the rest of the paper doesn’t hold up.

    Let me say that I think these are both good papers, but they are essentially starting with two different hypotheses about AdS-CFT, and I don’t see how they can both be applicable at the same time. The fact that nobody seems to have even noticed this is very troubling to me. But maybe I’m too much of a mathematician.

  26. Amitabh Lath says:

    Peter V, I understand your point but the decisions made by these top tier students does much more to sway these “people who might have something to say about whether string theory research gets supported” than some national lab’s public outreach ‘zine.

    Every grad program wants these students: sky-high physics-GRE, letters dripping with superlatives, transcripts with half a dozen graduate level courses completed as undergrad. They are courted with fellowships and awards. Their eagerness to join the field is seen as proof of vibrancy. If a big-name string theorist leaves your department and the acceptance rate for these blue-chips drops, you know the search committee will form quickly.

  27. Mark says:

    Peter Shor:

    The Shenker-Stanford butterfly effect has since been independently found in CFT using strictly field theory tools, so I’d say that it’s on a firmer ground than the error correcting code proposal. The HaPPY code’s claim to fame is that you can act on it with operators in the “bulk”, which is equivalent to acting on some “boundary” subregion. People don’t know how to do time evolution in this approach, so they can’t reproduce the butterfly effect from quantum error correction. But if I understood your comment, you’re saying that already acting on the state should take you out of the code.

  28. Blake Stacey says:

    Peter Shor asked, “[W]hy are so many physicists convinced that AdS-CFT shows that quantum gravity in our universe has to be unitary?” I have also wondered about this. AdS is said to be “a box to put gravity in”, but that comes at a price: The stability properties of gravitation on AdS are qualitatively different from those for dS or Minkowski backgrounds. The answer to the question “Will doing this experiment make my laboratory collapse into a black hole?” is, in principle, different, precisely because of that convenient boundary that makes for a nice box and gives a place for the CFT to live. This makes it hard for me to accept the reassurance that I was given many years ago, the standard colloquium line that we can put any local physics we want into an AdS box and apply the lessons from that exercise to our own world.

  29. Peter Woit says:

    I fear this is the wrong place to debate the issues raised by Peter Shor about CFT and error-correcting codes, partly because the moderator knows nothing about the topic (he would like to someday understand what that’s about, but today is not the day…)

  30. Peter Shor says:


    Just to be clear, I have no objections to the Shenkar-Stanford butterfly effect in CFTs that aren’t also error correcting codes.

    And probably we should stop any discussion here, to keep the moderator happy.

  31. niels abel says:

    CERN has a jpg and a pdf on today’s (2019-04-28) Header Image:

    From http://library.cern/archives/history_CERN/historical_images?page=2

    Pauli thought Heisenberg’s ‘World Formula’ needed a lot more work, and he made his point graphically. He sent this drawing of an empty picture frame to George Gamow on 1 March 1958 with the caption, ‘This is to show the world that I can paint like Titian … Only technical details are missing.’



  32. Pingback: The Scientific Attitude | Not Even Wrong

Comments are closed.