What Scientific Idea is Ready for Retirement?

Every year John Brockman’s Edge web-site hosts responses to a different question. This year the question was What scientific idea is ready for retirement?. It shouldn’t be too hard to guess what I chose to write about, with results available here.

Every year Brockman manages to attract more responses, so this is now getting to be a statistically significant sampling drawn from the population of people who write about science for the general public. Before trying to divine some general trends among the physics responses, I’ll first mention a few of them that stand out as unusual.

First, there’s one from Paul Steinhardt that I very much agree with. He’s had it with the multiverse and thinks it needs to go. I’m very glad to see someone else making many of the points that I endless repeat on this blog in a tiresome way. So, go read what he has to say, which ends with this challenge to the theoretical physics community:

I think a priority for theorists today is to determine if inflation and string theory can be saved from devolving into a Theory of Anything and, if not, seek new ideas to replace them. Because an unfalsifiable Theory of Anything creates unfair competition for real scientific theories, leaders in the field can play an important role by speaking out—making it clear that Anything is not acceptable—to encourage talented young scientists to rise up and meet the challenge.

It would be great to see someone other than him and David Gross start publicly speaking out.

A second outlier is Gordon Kane, who uses this as an opportunity to claim that he had predicted the Higgs mass using string theory. I don’t know of anyone other than him who takes this seriously. He doesn’t mention his other string theory based predictions, which include the prediction that the LHC should already have seen gluinos.

Another odd one is from Max Tegmark, who argues that we have to get rid of equations in physics that aren’t just based on finite and discrete quantities. The only positive argument I can see from him for this is that it would help get rid of the “measure problem” of the multiverse, but listening to Steinhardt and dumping the multiverse itself seems to me a much better idea. Tegmark has a new book out, I’ll write more about this here in a few days.

Maria Spiropulu is with me on the need to retire naturalness, also wants space-time to go. Getting rid of space-time has multiple proponents, including also Steve Giddings and Carlo Rovelli.

Another theme is people starting to sound like John Horgan, announcing we’re reaching the limits of science. Martin Rees thinks that some scientific problems may never yield to our understanding: “The human intellect may hit the buffers”. Ed Regis thinks the cost of a next generation collider is just not worth it for what it is likely to tell us.

A variant of this is the argument that we’ve reached the end of the road for unification and simplicity in our basic physical laws. Here the argument often seems to be that since SUSY/GUTs/string theory were such beautiful elegant ideas, their failure means the whole elegance thing is misguided. Another point of view (which I think someone wrote a book about) would be that these always were heavily oversold as “elegant”, since if you looked into them they were rather complicated and didn’t explain much. Writing in the anti-elegance vein are experimentalist Sarah Demers:

It is time for us to admit that some of the models we have been chasing from our brilliant theory colleagues might actually be (gorgeous) Hail Mary passes to the universe.

along with Marcelo Gleiser and Gregory Benford. At this particular time in intellectual history, it seems that hardly anyone has anything good to say about mathematical elegance as a powerful principle behind deep ideas about physics.

Finally, the biggest contingent are the multiverse maniacs. There’s Andrei Linde, who deals with the problem of evidence for his ideas by:

A pessimist would argue that since we do not see other parts of the universe, we cannot prove that this picture is correct. An optimist, on the other hand, may counter that we can never disprove this picture either, because its main assumption is that other “universes” are far away from us.

He’s joined by Sean Carroll, who wants to do away with the Popperazi and their inconvenient demands for falsifiable predictions. Also writing in support of the idea of a multiverse of different physical laws, implying we’ll have to give up on the idea of understanding more about the ones we see are Lawrence Krauss and Seth Lloyd.

Update: A couple more late additions that I missed. Eric Weinstein is with me in going after “string theory is the only game in town” as something that should have been retired long ago. Alan Guth uses this venue to promote some recent speculative work on the arrow of time with Sean Carroll (no paper yet, so hard to tell what it really is).

Update: Sean Carroll has a blog posting up about his argument for getting rid of falsifiability. He seems to not be getting a lot of support, either in his comment section (see for instance here), or places like here. I don’t think the skeptic community is ready to disarm itself intellectually in arguments against religious believers by ditching the conventional scientific method.

Update: Scott Aaronson writes here about the falsifiability issue, pointing out about string theory/multiverse research that

I wouldn’t know how to answer a layperson who asked why that wasn’t exactly the sort of thing Sir Karl was worried about, and for good reason.

Sean Carroll responds that the problem here is

somber pronouncements about non-falsifiability from fuddy-duddies.

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

70 Responses to What Scientific Idea is Ready for Retirement?

  1. Dom says:

    Does Carroll have an eye on The Templeton Prize?
    I’m sure they would love the implications for claiming any belief is as strong as any other regardless of falsifiability.

  2. philh says:

    I used to think falsifiability was the be and end all. However I am not so sure, maybe someone can bring me back, although I certainly don’t think it should be retired.
    To me the hypothesis that life has existed on another planet is a) not falsifiable and b) is a scientifically plausible hypothesis.
    There are scientific reasons to believe this hypothesis is true, and there has been scientific progress to making us more confident in its truth. The more we discover organisms that live in harsher and harsher environments there higher our confidence becomes in the hypothesis. Of course until we actually find life on another planet we should not be certain. But the way I see it we can never falsify the idea that there is no life on other planets. . Even if we visited every planet in the galaxy, life could have arisen on some of them in the time it took us to visit them, or maybe it went extinct before we got there. , or maybe it only arises on average in 1 in a million galaxies.
    To me science is a mix of theory and experiment, but with experiment being the supreme judge. A hypothesis that is falsifiable if far more preferable than one that is not. However in the real world it seems things are messier than just the issue of falsifiability.
    How could you falsify the idea that a human being could not survive a trip the centre of a black hole? But do you think such a statement has nothing to do with science?
    So whilst falsifiability is ideal, there seems to me to be other considerations. Is a hypothesis consistent with established physics? Is it implied by established physics? Can it ever be verified by data? These may not be as good a criteria as falsifiability, but i don’t think such consideration should be equivalent to religious fantasy.

  3. philh says:

    Sorry obviously the above should read you cannot falsify the idea that there is or has been life on other planets.

  4. I fully support emile’s statement about the quality of most scientific research in HEP and other fields of physics and astronomy. We do not need to obsess about the hype and we should not forget that much of scientific progress is incremental and unglamorous. The people who work hard for a modest advance should not be swept aside while we go after those who would occupy center stage. As George Ellis pointed out, there are serious and successful attempts to describe real progress in physics. What does it matter if uneducated undergraduates have heard more in the news about string theory than about the baryon asymmetry? Galileo, Newton, Maxwell, Lagrange, Heisenberg, Einstein, Zwicky and many others certainly did not worry about the hype of their days nor about what students were attracted to. There were plenty of charlatans in the past but physics and mathematic marched on. I hope it will continue to do so, despite some of the anti-science ideas expressed by notable, influential people.

  5. Peter Woit says:

    Falsifiability is definitely not the be all and end all of distinguishing science from non-science. Besides the example you give, there are plenty of others where one can see that this is a very tricky question. In situations where you don’t have falsifiable predictions you need to carefully think about what the nature of the evidence is for your scientific claims. The danger with arguing for “retiring falsifiability” is that you need to say what you intend to replace it with, and ensure that you don’t start adopting justifications for your ideas that are pseudo-scientific, applying equally well to religion, for instance. Some of Sean’s commenters are pointing out to him precisely this problem, that his argument for what is a scientific claim applies equally well to claims about the existence of a deity.

    In the case of multiverse theories, without falsifiability, you need to check to see what exactly is being used to justify them. Sean’s argument for the multiverse has often been that “String theory predicts it”, changing the problem to what the justification for believing in string theory is. Then you run into the argument that “we can’t test string theory because it predicts a multiverse, and we could be anywhere in this”. The problem here is that you have hit circularity.

  6. Low Math, Meekly Interacting says:


    I like to use “testable” as a criterion, if “falsifiable” is too strict or otherwise unworkable. I’m not well-versed enough in statistics to claim to be a “frequentist” or a “bayesian”, but the data we have on known life that you mention leads me (and many others) to conclude that testing exobiological hypotheses is an eminently reasonable goal. It is entirely plausible that technology will advance enough even in my lifetime to directly image thousands of exoplanets in our galaxy, and spectroscopic and other data derived from those observations is likely to be informative, perhaps even definitive on an individual basis. We can adjust our exobiological expectations based on those data. We’ll never be able to image all the planets around all the stars in all the observable universe, but we could reasonably hope to observe a fairly representative sampling. Surely we could make some very informed deductions once we have those data.

    Can we aspire to anything remotely like this example as we explore the hypothetical anthropic landscape, especially if it says nothing sufficiently specific about our vacuum to assess plausibility, and the remaining (effectively) infinite vacua are forever beyond our horizon?

  7. Chris W. says:

    The importance of some sort of falsifiability can be reduced to a simple question one can ask about any idea: “If this idea was wrong, how would I know it?” Whether observations and experimental results in the physical world are relevant to answering this question depends on what the idea is about, i.e., on what problem it was proposed to address or solve.

    Some people appear to conclude that not having an answer to this question justifies accepting the idea on faith, or according to one’s taste. Of course, one doesn’t have to fully accept an idea to investigate it and develop it to see where it leads, but the endgame in science is finding a way to meaningfully check it against observation.

    I should emphasize the asymmetry that Popper wrote about frequently. Having a way to check an idea of universal scope—as physical theories are—in no way allows one to prove that the idea is correct. At most, it allows one to establish that idea has value, insofar as it has passed many tests and has (perhaps) been effective in applications.

  8. Dom says:

    Excuse me if I have this wrong but if we negate the idea that Life exists or has existed on other planets don’t we falsify it by finding evidence that it does or has? Isn’t the framing of the original position the problem here?
    Using appropriate language we can easily create apparent problems with logic e.g. “This statement is false”

  9. JR says:

    I am with Dom on this. “life has existed on another planet” is not falsifiable, but the converse: “life has never existed on another planet” is definitely falsifiable. So we can, in principle and in practice, make progress. Compare this to “The multiverse exists” and its converse “the multiverse does not exist”. As far as we know right now, neither of these is falsifiable, either in principle or in practice, so they are outside the scope of science.

  10. imho says:

    philh said:
    To me the hypothesis that life has existed on another planet is a) not falsifiable and b) is a scientifically plausible hypothesis.

    I don’t think this makes any sense. Using your logic, physics couldn’t make the statement “all electrons have mass X”, or “Every action has an equal and opposite reaction”. In this universe, there is no such thing as “absolute falsifiability” as you seem to be suggesting. I have no idea about all this Popperian pontificating, but the operational real-world criterion for scientific acceptability is: overwhelming evidence in-support-of with no know counter evidence against. For example, we can never measure the mass of every electron in the universe (even in principal), but in the 100,000 experiments that have been performed that could detect such a thing, we have never seen a deviation from the accepted quantity. Also, the idea that all electrons have the same mass fits perfectly into the currently accepted wider Physics framework. This discussion needs less philosophical nonsense and more “shut up an calculate”.

  11. Peter Woit says:

    Maybe that’s enough about falsifiability for now. I don’t think it’s a topic that lends itself to this kind of unfocused discussion.

  12. Simple biologist says:

    Reading the Edge pieces by theoretical physicist on Edge is doing wonders on getting rid of whatever residual physics envy I had left.

    Thanks for Carrol et al. for that.

  13. theoreticalminimum says:

    I would like to react to Tegmark’s response. When I was an undergraduate, it really struck me when I learnt that Nature realises the (regularised) infinite sum 1+2+3+… = -1/12 in the Casimir effect; one computes the Casimir force as being proportional to this sum, and the result is exactly what one measures in the lab. Wouldn’t this stand as a counterargument to Tegmark’s comment about retiring the idea of infinity? Without the series being infinite (but regularised), we cannot compute the Casimir force.

  14. Chris W. says:

    This discussion needs less philosophical nonsense and more “shut up and calculate”.

    Don’t forget “shut up and do some experiments” and “shut up and think”. These days there is plenty of calculating going on, but not necessarily with much point to it.

  15. Shantanu says:

    Peter, something OT to this post, but not OT to the blog.
    video to a seminar by Paul Langacker at PI which gives an idea
    of where particle physics/string theory is going.

  16. Peter Woit says:

    Thanks. Looks like the story is still going to be “string theory very promising”, no matter how clear it becomes that it predicts nothing and the LHC sees nothing even vaguely encouraging.

  17. philh says:

    Hello Peter, thanks for your for your reply. I’m glad you agree that falsifiability is not the “be all and end all” of science.
    I also agree that it should not retired in favor of a theory of anything,
    One difficulty I have is to class all multiverse theories in the same set.
    Like you, I’m unimpressed with the string theory landscape, string theory has not had any empirical data to back it up, as you point out there has no super symmetry, extra dimensions or black holes found at the LHC. I’m also unimpressed with the many worlds interpretation of QM; there are too many interpretations and no data to choose between them.
    However I find it had to put the inflationary multiverse in the same bracket. The reason is just as experimentalists have failed to find anything like what strong theorists hoped for at the LHC, experimentalists have said the data favors inflation. Wmap and Planck team said this.
    At the same time the source you quoted as the one you agreed with (Paul Steinhardt) said
    Inflation leads to eternal inflation. Guth said the same thing..
    I’m not qualified to say if Guth and Steinhardt’s statements are true or not. But If I see theorists telling me inflation is generically eternal and experimentalists telling me the data is favorable to inflation then I have to say the inflationary multiverse does not appear to be in the same boat as other multiverse proposals like the string theory landscape. Steinhardt also said it would be possible to falsify inflation by detecting a blue tilted gravity wave spectrum.
    So whilst the generic idea of a mulitverse may not be falsifiable , the specific idea of an inflaitonary multiverse could it seems be falsified simply by falsifying inflation in the way the Steinhardt has suggested.

    Perhaps the problem is for people to decide if a scientific hypothesis is true or false or science or not science. Why can’t each of these be a continuum? With some things obviously false and obviously pseudo science, other things being a plausible scientific hypothesis, and of course, yet others being confirmed solid science.
    So If Sean Carroll tells us there’s a multiverse because of the string theory landscape I think he’s on vary shaky ground, but if says it on the back of the inflationary multiverse then it seems to me to be a little bit more solid. Have I gone wrong here Peter, if so where?

  18. philh says:

    Hellow IMho, your discussion of electrons is interesting. But there is a big difference between saying the electrons has mass x and answering the question, is there life on other planets?
    The difference is staring you in the face, we have electrons and we can measure their mass. It’s not the same with exo planets. We cannot answer the question of life on other worlds with “shut up and calculate”. How are you going to do that?
    Dom and Chris please note my discussion of multiverse above.
    Low math I agree testable is better than falsifiable. However whilst I agree the tech will get better to characterise exo planets. If we find nothing, that wont falsifies the idea of life on other planets. Under the falsifiable framework people looking for exo planets aren’t doing science, but under the testable framework they are. That’s why I prefer the testable framework as better. Although a theory that is falsifiable will be always be more compelling than one that is not. Again I’m calling for a continuum here rather than black or white criteria.

  19. Kavanna says:

    String theory doesn’t predict the multiverse. String theory predicts nothing.

    The multiverse idea was cooked up precisely to *prevent* string theory from predicting something, something clearly wrong, and thus be disproved. The multiverse idea gained ground to keep the string party going. Attempts to get rid of falsifiability are motivated by the same motive. It’s a racket.

  20. Low Math, Meekly Interacting says:


    I think we’re largely in agreement. Most of the arguments against “falsifiability” are straw-man nonsense anyhow, but occasionally one has enough nuance to worry about. Absolutists aren’t helpful in any case, which is where the “probabilistic” moderation I think is helpful. I doubt Popper himself would have argued otherwise. You’re right: falsifiability is always highly desirable. But if I can never falsify the prediction that “some day the Sun will rise in the West”, I can make a pretty compelling case for summarily dismissing such foolishness on time scales any mortal ought to be worrying about.