Only a theory?

November 15, 2014 • 8:52 am

Matthew Cobb sent a link to this video about the nature of a scientific theory, along with the note:

Brief 3-minute video on theories in science from the Royal Institution, written by science teacher and atheist Alom Shaha and narrated by theoretical physicist and humanist, Jim al-Khalili

Nothing you haven’t said a million times over, but nicely done.

Indeed, but you can’t say it too often. The notion that evolution is “only a theory,” with “theory” construed as “a wild guess or an unevidenced speculation” is perhaps the greatest public misconception about evolution. Save this video to show to those doubters:

I also believe that theories can be “truths” if there’s sufficient evidence supporting them that, as Steve Gould said, it would be perverse to say they’re wrong.  Remember that we still have the “atomic theory” and the “germ theory” for the constituents of elements and the agents of infectious disease. To say that the “germ theory” isn’t “true” seems wrong to me.

One question: the third criterion for a theory given in the video is this, which they consider the most important: “a scientific theory provides ways to make predictions about the aspects of the world it explains, which we can then test by further explanation.”

Does that meant that string theory, since nobody’s found a way to test it, isn’t a scientific theory?

87 thoughts on “Only a theory?

  1. I’d say string theory was a ‘theory in waiting’ as it is not inconceivable that we might devise ways to test it in the future. I dimly recollect reading once that the theory which said the universe might be some kind of hologram was untestable, but later someone devised a clever way to test it. I’m sure there must be other examples. Other theories might be untestable in a more profound and intractable way, so I guess they’d have to be classified as speculation. But it would be a brave scientist who asserted that a particular theory would always remain untestable.

    1. Just as important as it not being ‘inconceivable that we might devise ways to test it in the future’ is the attitude that this would be a good thing. In fact, making testable claims is a very important, significant, crucial thing in science, and advocates of String Theory are all desperately hoping and working so that some day this will come about.

      Contrast that with the attitude of religious faith and conspiracy ‘theories,’ where the lack of testability is considered a feature, not a bug. The people who believe in it anyway are exhibiting humility, or open-mindedness, or super-cleverness, or love. They have distinguished themselves from the hoi polloi and their need for “proof.”

      I have had religious apologists compare belief in God to belief in String Theory, acting as if they have now granted themselves respectability if not sympathy. But of course this difference in attitude and approach is a major distinction between the two. Instead of struggling to find ways to devise testable predictions for the existence of God so that it might be placed more firmly on an objective level, they really just spend more time looking for analogies for why they shouldn’t have to try to treat their faith beliefs as objective claims. The String Theory analogy is supposed to be a tu quoque. Major fail.

      1. Good point. Contrasting string theory with religion is useful. When that is done, it is clear that if anything, string theory has made every attempt to be a scientific theory. It’s not like it gave up, which is what religion does.

    2. String theory is “waiting” in the sense that it isn’t much useful for observations as of now.

      But that is different from a theory being testable. String theory is a good example, since it is already tested on certain predictions and since it is known to be testable at Planck scales where properties of any string would be seen in detail.

    3. Science has theories but only maths produces theorems . Only maths has ‘proof’ in that sense.

      Scientific theories are provisional; that’s a strength, not a flaw.

      And while Darwin’ theory was broadly correct, it was incomplete. Genetics has made it more complete; there may be more to come – though that ‘more’ doesn’t entail supernatural intervention.

      So while it hasn’t been ‘proven’ in the sense we’d use that term when discussing Fermat’s last theorem, it has been proven to an extent it would be unreasonable to disbelieve it but not unreasonable to believe there may be more to come.

  2. String theory definitely is not a theory in the strong scientific sense. I wish the term had never been invented. There are other plausible scientific hypotheses that aren’t identified as “theories”, such as supersymmetry and loop quantum gravity (although Wikipedia repeated call LQG an theory, which it shouldn’t IMO).

    1. What do you mean by “strong” sense?

      A theory is testable or not. String theory is testable. Then you can wish it was useful for more observations, but, alas…

      I don’t find loop quantum gravity a scientific theory, at least as it was intended to be. It is motivated by math (while string theory is motivated by physics) and it has failed to show a dynamics. You can’t make an energy (with a lower bound) and a harmonic oscillator in it, so it isn’t fit to do anything in physics. Physicists that aren’t in the fringe seem to accept it as a good faith effort, but little else.

      That it will ever do something in physics is not plausible, so I guess we will have to differ in this.

        1. It’s a theory in the same sense as number theory or set theory, i.e. a coherent body of theorems about a particular class of mathematical objects. Whether those objects accurately describe anything in nature is a separate question, but a negative answer wouldn’t make string theory any less of a theory in the mathematical sense.

            1. Well, the fact is that working physicists call it string theory. So I guess they’re OK with the mathematical sense of the word.

    2. ST seems to fit one part of the definition of a theory as an “explanatory framework” but the part about testable predictions it does not fit quite as well.

      1. Although isn’t it more about testable predictions that would distinguish it as being an improvement over other theories? I would be amazed if anyone were pursuing it if it were not capable of modelling reality as well as other theories.

    3. It would be correct to refer to the strings hypothesis. As we sit here today, strings may best be described as the strings hypothesis. As I understand it, strings is a branch of mathematics which may or may not be applicable to physics. It has a number of interesting properties, including possible applicability to gravity, but currently is unsupported by evidence. This is to be contrasted with the theory of evolution, relativity, and quantum mechanics which are strongly supported by evidence.

      Thus the answer to Prof. Coyne’s query is that, as we sit here today, strings is not a theory in the same sense as those cited above, but is a hypothesis.

      1. We don’t talk about “number hypothesis” or “set hypothesis”. Mathematical theories are still theories whether or not they happen to apply to physical reality.

        1. Excuse me, when I referred to the mathematical branch of groups on Panda’s Thumb as group theory, a mathematician took exception to my referring to groups as a theory of mathematics in a comment in response to my comment. Apparently, at least some mathematicians don’t like branches of mathematics referred to as theories of mathematics. Group theory as applied to physics is a theory of physics, not mathematics.

          1. The mathematicians I know don’t seem to have a problem with calling it group theory, and Wikipedia seems to concur. I’d be curious to know what your dissenter thinks it ought to be called instead.

    4. My problem is that the use of “theory” in this case is invalid for what it means in science.

      So shouldn’t it be “Super String Hypothesis” and not “Super String Theory”? To me it muddies the waters even more on meaning use an misuse of the meaning of theory.

  3. This is an excellent video.

    While “it’s just a theory” is popular (on Twitter, my playground), “If we came from apes, why are there still apes?” is a bigger problem. When I tell people that we didn’t come “from” apes but that we ARE apes, well, that doesn’t go over too well. (When people use the word “monkeys” in the question instead of “apes” I usually say: “We didn’t come from monkeys but DNA proves that monkeys and bonobos and gorillas are our cousins.” That usually enrages a few people as well.)

    So I’d like to see a comparable helpful video by someone who contends with this apes/monkey question.

    1. It can take time to address common misconceptions like that. It may be helpful to first ask them: ‘In what part of the world were your ancestors living a couple centuries ago?’ If they answer for example ‘Germany’, then my immediate response is ‘then why are there still Germans’?

        1. Chihuahuas were bred from red foxes. But the red wolves and foxes were natural to the region, chihuahuas were not. You really don’t understand evolution at all with that piker question. And you wonder why those who spend time on this get first bothered, then angry with the same wrong questions by those who know little or nothing about evolution ask them usually to take down Evolution and replace with with some version of Creationism.

          How about you do some studying so you won’t start on the wrong side by asking ignorant questions. Ignorance can be solved with education. I try to educate myself all of the time because we are all ignorant in one form or fashion of many things. So it isn’t an insult–it is an admission you need education. So go do it then return with better questions.

          Evolution is a broad part of the biological sciences. Tell us how Creationism can fit since that is something that is rarely brought up by Creationists who only want to knock down Evolution. I dare say they can’t be scientific and just speak of how Creationism fits as well as Evolution does in the biological sciences.

  4. String theory is remarkably distinct in that it utilizes quantum field theory, among other theories and tries to explain something more fundamental about nature.

    I can think of no theory in science that tries desperately to link itself to the observable world. It’s like a thoery that undermines itself, because it stretched too far to be tested…but it has no choice. My guess is that any thing that accomplishes more will also be plagued by the same problems of testability or have the incomprehensible feature of overcoming the burden of un-falsifiability.

    1. I have felt that although string theory has merit to it, we are unfortunately required to stretch our definitions in order to consider it a ‘theory’. Many people will unkindly call string theory a ‘religion’, since it too has no evidence, and yet holds many committed adherents.
      Maybe we should have a different term for a big-picture scientific idea that has explanatory power and yet lacks forseeable means of testing. ‘The string conjecture’, maybe. That sounds like an episode of the Big Bang t.v. show, but so be it.

      1. It is wrong to say string theory has no evidence.

        Now people can vary in judgement over exactly what is to be considered evidence and what it has as support.

        But there are people who tie evidence to string theory, and it isn’t exactly hard. (Consistent with current theory, useful for math/simplifications/toy model understanding, predicts nucleon flux tubes, black hole entropy, supersymmetry which is the last empty slot in semi-classical physics after mass (Higgs) was taken, predicts a smooth spacetime as seen in supernova photon timing and polarization observations, …)

    2. It isn’t a theory, it is a hypothesis in search of being a theory. Call it that and you will see why it really isn’t proven yet to become a theory.

  5. Never understood the opposition to starting a new convention for the capitalization of the “T” for a scientific Theory, to help the distinction with a mere layman’s conjecture.

    Save a lot of aggravation.

  6. Does that meant that string theory, since nobody’s found a way to test it, isn’t a scientific theory?

    In the formal sense of the word, it’s not a scientific theory. It will still be called a theory in the colloquial sense however, and I think that’s OK.

    1. I am all for ending the colloquial use of theory since it is wrong and causes no end of confusion. End it!

  7. Saying that “germ theory” isn’t “true” seems wrong mainly because “germ theory” does an awfully good job of explaining observations and predicting outcomes.

    Like Milton Friedman explains (and I’m paraphrasing): The theory that best predicts the play of pool players is physics. That doesn’t mean that pool players are actually making momentum/force calculations on the fly. But it APPEARS as if they do. And they may! We’ll never know. But if physics could predict billiards play with 100% accuracy, then what difference does knowing the “truth” make? There may be some alternate explanation for how billiards players make decisions, but that “truth” would be indistinguishable (in the data) from a “false” theory that has 100% predictive success.

    String theory is kind of a bastard child between physics and mathematics. It is a mathematical model that can explain certain peculiar aspects of the real world, but as you point out, it cannot be tested. How then is it different from the several Creationist “theories” we all mock? You have no choice but to accept in on faith.

    But at any rate, to hold “scientific theories” to such a high standard — and to label untestable theories as what they are: invalid hypotheses — isn’t obstructionist! It’s a discipline to which we owe for so much progress.

  8. Does that meant that string theory, since nobody’s found a way to test it, isn’t a scientific theory?


    But it is a set of (as yet untested) predictions based on quantum theory, . So you could say it is a (not yet scientific) sub-theory.

    Oh, and thank you, Jerry for sending me the article about Hitler and “survival of the fittest”. You’re a mensch.

    1. Technically, it is the String Hypothesis in the context of science. However, it might be possible to think of it as a theory in a mathematical sense i.e. there might be some basic assumptions and some mathematical theorems deduced from the assumptions. This would make it a mathematical theory à la group theory and set theory.

      I use the word “might” because I have only the vaguest conceptual understanding of what string theory is.

  9. Yeah, I’d say that the “theory” in “string theory” is being used much in the same way that the word “theory” is being used in the expression “music theory”. It’s like using the word to mean a process, based on some framework or construct.

    It’s really a simple conjecture (fundamental bits of stuff can be conceived of as 2D objects: strings) which initially held promise as a way of getting the maths to work so that we could unify gravity with the rest of the Standard Model. From what I’ve read of the early mail exchanges in the field, the initial expectation, given how well things worked out on paper, was that there would be some loose ends to tidy up, but in about 6 months time, humanity would have its Grand Unification Theory in place, and that would be that.

    Forty+ years later, the conjecture seems only to have spawned 10^500 or so possible avenues of investigation, indistinguishable from one another without experimental evidence seemingly impossible to obtain without fantastic energies we’ll never be able to achieve. And the main critics (who I tend to side with) maintain that, while this work was all worth doing, other avenues have been sorely neglected because of the cutthroat way ideas are defended and others derided in academia.

    1. Sorry, but opinion isn’t a valid judge of a scientific theory.

      As for unification of gravity with the rest of the standard particles, that is what GR did. You can predict a graviton by quantizing the low energy, large scale linear GR. All string theory contributes, as I understand it, is to predict gravitons also in the high energy, small scale region. But then you don’t really know if the Standard Model go that high in energy instead, you yourself mention GUT.

      For the last paragraph I like to point out that there hasn’t been any worthwhile contenders to string theory yet. (Which is a problem rather than a feature in the early days of a theory.)

  10. That is a good video, but it is almost too fast for me. The brisk, jazzy music is almost in the way. Still, it is one to keep around, IMO, and so I have bookmarked it.
    I have really liked the S.J. Gould essay: Evolution as Fact and Theory. This article is ‘up there’ in my personal list of great science essays.

  11. In a closely related vein…since yesterday’s discussion, I’ve been thinking more about science and proofs.

    The common tropes, of course, are that, “Science never proves anything,” and, “You can’t prove a negative.” But I would argue that that’s exactly bass-ackwards.

    Science is superlative at proving negatives, such that that’s basically all it does. That’s exactly what, “falsification,” means.

    Michelson and Morley inadvertently yet most emphatically proved that the Luminiferous Aether simply doesn’t exist.

    Today’s physicists have already proven that neither Quantum Mechanics nor Relativistic Mechanics is complete, even at the same time that they’ve proven that nothing else exists at relevant scales.

    Jerry, this Web site’s eponymous book lays out the proof that competing theories of the origins and development of life on Earth are false. As with physics, we know that there’s still to learn…but, again, at relevant scales, Evolution is all there is.

    So, the next time somebody lets loose with either trope, let ’em have it with both barrels.

    Science proves negatives.



    1. Yep. “You can’t prove a negative” has always rankled me. Another rankler for me is: “the plural of ‘anecdote’ is not ‘data'”. When actually, it can be precisely that — anecdotal data. My field is rife with it. Some bad, but some really, really good. It all depends on the methods.

      1. Yup. Conduct a sound survey to collect anecdotes, and, at the least, you’ve got data on the prevalence of said anecdotes.

        If you’re an epidemiologist and every patient with a certain set of symptoms comes in with a similar anecdote, if you don’t consider that data you’re failing at public outreach at the least.

        I’m sure you’ve got much better examples than any I could think up….


        1. Here’s a big one: selective dismissal of the anecdotes of HIV-positive women in sub-Saharan Africa, when they maintain they’ve only had one sex partner (i.e., calling them liars to maintain the fiction that 90% of HIV is penile-vaginal worldwide, a conjecture with no real scientific basis beyond ecological-level evidence). …while not investigating clustering of cases in the prenatal clinics they attend where seven syringes, rinsed in tap water are used to inject a day’s typical group of 300 women. I’ve got a zillion more in this… uh… vein, which – when taken together – raises serious concerns for how epi is conducted. But to come to these conclusions, one must first take such anecdotes seriously enough to begin the investigation. (i.e. contact-trace & conduct genetic studies, none of which has been done there). I won’t derail further here, but that’s the biggie being swept under the rug in my field. The Ebola madness was a continuation of similar money and power-grubbing machinations – by well-meaning people, no less.

    2. The problem is that most terms are deepities and spiritual thought processes always trade in on deepities.

      Of course we can prove a negative — unless by “prove” you mean 100% exclusion of any logically possible alternatives. Which is what the ‘gotcha!’ folks usually do mean (you can’t prove a universal negative) — until they get off their favorite protected belief and join the real world and start speaking reasonably again.

      They turn it off and on. I have friends who actually argue that well-attested scientific theories are technically only opinions — matters of faith — because they’re starting out with the belief/knowledge that the physical world is an illusion created by an excess of fear and arrogance and a lack of love and acceptance. Yes, really. I mean yes, really, this is their argument. “Prove” it wrong.

      Pointing out inconsistencies in this (they’re quick enough to accept scientific results when it suits them) is tricky, since they also think logical thinking is judgmental thinking, and the sign of a person who just wants to condemn things they don’t understand. Love conquers all — including contradictions.

      So yes, your point is a good one. But sometimes it’s still an uphill battle even when you’re well-armed.

      1. Of course we can prove a negative — unless by “prove” you mean 100% exclusion of any logically possible alternatives. Which is what the ‘gotcha!’ folks usually do mean (you can’t prove a universal negative) — until they get off their favorite protected belief and join the real world and start speaking reasonably again.

        To the former, there’s always an infinite variety of conspiracy theories that can explain why something exists but you don’t observe it. Brains in vats, subroutines in the Matrix, alien mind-control rays — the list really is endless. And the very textbook definition of insane.

        To the latter…certain universal negatives are the easiest of all to demonstrate, as well as the oldest examples of “proof” in any sense of the word. There is no “largest prime number,” for example, most famously. And, pace Epicurus, there are no powerful agents with humanity’s best interests at heart. A single example of evil that such an agent could have prevented but didn’t is sufficient to demonstrate boundaries on the hypothetical agent’s scope — and the most popular proposed agents allegedly are limitless. Every time an human official of the gods buggers another child and the gods fail to such a trivial thing as call 9-1-1 in response, the incompetence and / or malevolence is again demonstrated. Such can be explained away, of course, but only by agreeing with the disproof of the universal claim and severely diminishing the local claim. Maybe there are gods, but they’re nowhere near here and / or they know nothing of us and / or they’re utterly powerless and / or they hate us. But why, then, call them gods?



    3. Science (logic) is the only chance to prove a negative.

      The criteria for a scientific theory is actually easy:

      Can it predict the outcome of an event, like A -> B. Physics simplifies this prediction to:


      Ironically the foundations of physics is pretty much making zero smaller, searches for deviations to G, EDM, etc. Religion never comes close.

  12. The evidence and reliability of a theory certainly changes over time. String theory will likely remain on the edge of a scientific theory until more can be tested and known to either bring it further along or discard it.

    Same thing goes for denial of scientific theory. The people who were in denial of Evolution 100 years ago were not as seriously stupid as those today.

    1. I wish I had something to contribute here, but in the absence of that let me say that this captures the essence of it as I understand it.

      The chips may fall either way (and if string theory isn’t realized as physics, physicists have a huge problem), and we will have to wait and see.

  13. Theory, belief and faith: lots of words have multiple meanings, but I can’t think of any that result in more confusion and mischief. I find I say “I believe” less and less, and “I have faith” not at all. I probably contribute to the middle on the word theory, but luckily I don’t actually have any theories, so …

  14. I wouldn’t have done the video that way, but it is good.

    But some problems makes it perhaps not good enough in all cases. I am thinking of the hasty generalization that all theories will see a nice raise to a final success, which I’m sure people will point at and say “see, there is no difference with a lot of these and a lot of general theories”.

    I think I would have pointed out that in some cases there will result in only one theory, and that is the interesting cases. (Atomic theory, germ theory, …) That never happens for colloquial theories.

    Does that meant that string theory, since nobody’s found a way to test it, isn’t a scientific theory?

    Now you have listened to the opinionated nutters too much.

    Remember that “theory” here is originally used in the mathematical sense, since string theory results in unique and/or easily found math results. There is nothing wrong with that sense of theory either, and it is quite a confirmation of the video even if they forgot to use it.

    Moving on to the physics:

    String theory predicts the same flux tubes as QCD does and the same black hole entropy as QED + GR do. Those tests weren’t all just consistent with current theory either, string theory was a year quicker than QCD, it’s just that QCD is the simpler theory.

    And who knows? String theory is testable latest at Planck scales, and BICEP2 may still be correct in that inflation is shy 2 orders of magnitude. The day after the ill-fated BICEP2 press conference there were a scientific meeting where they discussed how string axion mechanisms were the only way to predict the seen field strengths…

    1. So, would it be fair to say that string theory (math[s] sense) is not yet a theory (science sense) of physics, only a well-validated hypothesis (or set of hypotheses)?

      So, for now, it’s the string-theory hypothesis; later, it may be the string-theory theory…


  15. I wonder if people aren’t being a bit too fastidious about the notion of theories as empirically confirmed hypotheses. Newton’s theory of gravitational is still a theory (isn’t it?) despite being experimentally disconfirmed with regard to the precession of Mercury’s orbit. That is, it’s a useful explanatory framework that makes accurate predictions most of the time. It just happens to be wrong about the actual physics.

    I’m sure we could name any number of other superceded theories that nobody bothers to call hypotheses. They’re theories that turned out to be wrong.

    On this view, string theory is indeed a theory, and the yet-to-be-confirmed hypothesis is that string theory describes reality.

    1. “Newton’s theory of gravitational is still a theory (isn’t it?) despite being experimentally disconfirmed with regard to the precession of Mercury’s orbit.”

      Newton’s theory of gravitation is a special case of general relativity. It wasn’t so much disconfirmed as superceded.

      1. Well, it’s not really superseded either. Unless you’re worried about GPS, Newton is still fine for everyday life. No-one would use GR to calculate the path of a tossed ball, for example.


        1. In the context of fundamental physics, it was superseded. There’s no clearer case, that I know of, of one correct but incomplete theory being overtaken by another consistent but far more general one.

          1. I think you’re being generous in calling Newtonian gravitational theory “correct but incomplete”, given that there are known cases (such as the orbit of Mercury) where it gives incorrect answers.

            1. That’s the sense in which it’s incomplete! As I indicated above it’s more-than-adequately correct at earthbound human mass/energy scales.

              If string theory becomes a fully-fledged theory of physics, do you think the Standard Model would no longer be correct?


              1. “Incomplete” would mean that there are cases where it makes no predictions at all, where the math breaks down or becomes intractable.

                Mercury’s orbit isn’t like that. The Newtonian math is clear; it just gives the wrong answer, where everyone fully expected it to give the right one. Einstein didn’t make Newtonian gravity incorrect; it was already known to be incorrect before Einstein formulated its replacement.

                Nor will string theory make the Standard Model incorrect. If there are known cases where the SM gives wrong answers (as distinct from no answer at all), then it’s already incorrect, independent of string theory.

                There’s nothing that says a theory can’t be incorrect in some details but still useful — unless you insist on a definition of “theory” that precludes any degree of incorrectness.

              2. Well, if you want to define “incomplete” in that way. 😝

                It is complete within the regime I noted elsewhere, which doesn’t include Mercury. It is incomplete inasmuch as it ignores second order effects (“gravity due to gravity”).

                SM is *known* complete and correct only up to Higgs energies. (And maybe not that; google ”techi-Higgs”.)



          2. Oh there’re plenty of examples. My own personal favorite is the shape of the Earth. Pretty much everybody but cartographers and intercontinental pilots spend their entire lives exclusively using the Flat Earth theory for navigation. And even the latter can get away with the actual maneuvering (not the initial course plot) by way of lots of course corrections to a flat model.


  16. At what point did the theory of evolution by natural selection &c. become a “theory”?

    At what point did the germ theory of disease become a “theory”?

    At what point did atomic theory become a “theory”?

    At what point did the heliocentric theory of the solar system become a “theory”? (Long after Galileo!)

    What criteria did these all meet at that point and who decided to confer that status?


Leave a Reply