Is falsifiability essential to science?

Why Evolution Is True

The two articles I want to discuss today are fascinating, for they raise a problem that’s now vexing many scientists (especially physicists)—the problem of testability. (Thanks to reader Mark H. for calling my attention to them.)

It all goes back to the philosopher Karl Popper (1902-1994). Popper’s views about what made a theory “scientific” were immensely influential. They’re summed up in the Wikipedia piece on him:

A theory in the empirical sciences can never be proven, but it can be falsified, meaning that it can and should be scrutinized by decisive experiments. If the outcome of an experiment contradicts the theory, one should refrain from ad hoc manoeuvres that evade the contradiction merely by making it less falsifiable.

In other words, a theory that can’t in principle be shown to be wrong isn’t a scientific theory. But I disagree with that characterization, and the one from Wikipedia, in two ways.  First, a…

View original post 2,389 more words

12 thoughts on “Is falsifiability essential to science?

  1. The problem in accepting a theory that is “beautiful” but cannot be tested is that it may simply be garbage. If it is, everything that is built on it is garbage too. So it would not be wise to accept such a theory.


    1. Totally agree. In one of the articles on this conference, someone quotes Einstein and how he judged his own theories by their elegance. It pays to remember that Einstein was initially dead wrong about the universe being timeless and eternal, and that he never accepted quantum mechanics due to its perceived ugliness. The universe has demonstrated over and over again that it has no obligation to conform to our sense of beauty or elegance.


      1. Absolutely. If we stop insisting on falsifiability we’ve returned to a pre-scientific age where we believe what we want to.

        Seems like I must quote Feynman here: “The first principle is that you must not fool yourself – and you are the easiest person to fool.”

        Liked by 2 people

  2. I confess I didn’t have time to read the full article, but elegance is something that I believe should be taken into account. I admit there’s more involved than elegance, and that this word should be elucidated. It’s not “beauty” in the sense of a walk through the woods on a clear autumn day.

    That said, I like this author’s description of falsifiability, which he’s made a bit looser than Popper’s. (Or rather, his interpretation of Popper. I haven’t read Popper in a long time and don’t remember the details of what he said about falsifiability.) I think it makes sense to say a theory should be conceivably falsifiable rather than demanding that it be falsifiable now.


    1. Words like “elegance” and “simplicity” can be tricky. For Occam’s razor, I like looking at how many assumptions a theory has to make. Each assumption is an opportunity to be wrong, and weakens the prospects that a speculative theory reflects reality.

      From what I’ve read, Popper’s criterion was that it only had to be falsifiable in principle. (Although I have to admit that I’ve never read Popper directly. My knowledge of his views come from philosophy of science overview books.) If it wasn’t even falsifiable in principle, in other words, if no one could conceive of any way it could ever be tested, his view was that it was metaphysics, which he still considered a worthy endeavor, noting that atomism was metaphysical speculation to the ancient Greeks.

      Myself, I’m okay with calling things like the multiverse and string theory speculative science, since they arise from mathematics that take established scientific theories as their starting point, as long as the “speculative” part is kept clear.

      Liked by 2 people

      1. That was what I remembered of Popper, and what I would’ve said if questioned about it, but it’s been so long. I started to wonder if I’d misread him.

        Agreed on “speculative.” There’s got to be something that defines science and delineates it from metaphysics (which, as you probably know, is not a word that bothers me…unless it refers to special crystals with healing powers). It seems that falsifiability is only one component of “science,” but it’s important.

        Liked by 1 person

        1. The demarcation between science and metaphysics is actually pretty hazy. When I read Jim Baggott’s ‘Farewell to Reality’, which is about this whole issue, he noted that just about every scientific theory is metaphysics, since they have elements that can never be confirmed, only falsified. He was okay with that, as long as the overall theory wasn’t only metaphysical. (Which in his view many speculative theories are.)

          Liked by 1 person

  3. I’m not a fan of theories that cannot be tested despite their putative mathematical rigor. I generally don’t consider those “theories” so much as “hypotheses” (and I’m fine with the latter).

    I do think an appreciation for elegance is useful, but — as with Occam’s razor — it’s just a heuristic, a guiding principle. But physics does tend to be elegant in the final analysis, and Einstein’s objections to QM were understandable. Certainly the current particle zoo is pretty messy.

    (Speaking of which, have you heard the rumblings about the statistical “bump” at CERN? It’s interesting in that it’s showing up in both the ATLAS and CMS experiments. Might we be seeing the first crack in the SM? God, I hope so!! XD )


    1. On hypotheses and theories, that’s a sentiment that I often hear, that an untested theory is a hypothesis. It’s one many people take to challenge the notion that theories like natural selection are tentative because they’re “just theories”. I totally understand that impulse, but I can’t see that it’s that clean.

      From what I’ve seen in actual scientific circles, hypotheses are usually tentative but relatively simple explanations for a phenomena. “Theory” usually refers to a more comprehensive model, which may be speculative or it may be well tested (or even falsified). I think that’s why string theory isn’t called string hypothesis. But it’s not something that is uniform across all academic fields. (Biblical criticism seems much more likely to use “hypothesis” as in the Documentary Hypothesis.)

      I did hear about the bump. From what I understand, it’s currently at 2-sigma confidence, so many physicists are remaining cautious. (Although it sounds like there are already a flurry of papers speculating about it.) I think most of the physics community would like it to just be something unexpected so that it has some sort of theoretical implications, just to shake things up.


      1. “…that an untested theory is a hypothesis.”

        Just to be clear, I said “cannot be tested” — which isn’t the same thing!

        My response to those who try to deny natural selection as “just a theory” is to suggest they accompany me to the top of a 20-story building, step off the edge, turn around to face me, and discuss the “theory” of gravity. >:D

        I expect it to be a short (and rather doppler-shifted) argument.

        “From what I’ve seen in actual scientific circles, hypotheses are usually tentative but relatively simple explanations for a phenomena.”

        Not all circles are the same. 🙂

        I’ll defer to Ronald H. Pine on this one:

        A theory, per Wiktionary, is: A coherent statement or set of ideas that explains observed facts or phenomena, or which sets out the laws and principles of something known or observed; a hypothesis confirmed by observation, experiment etc.

        In the physics community, a theory very specifically is a set of one or more equation(s) that exactly specify some behavior. For example, there are theories (that is, equations) behind each of the four forces.

        Kinda, sorta, the more words you use, and the less math, the more it’s hypothesis territory, not theory territory.

        “I think that’s why string theory isn’t called string hypothesis.”

        String Theory is a body of mathematics. (One that may — or may not! — have any connection to reality.) The idea that reality is made of strings is a hypothesis.

        Ultimately it’s a “how you define it” thing so if a conversation turns on those words, we’all just have to understand how we’all define them. Y’all.

        More importantly, is Trump a theory or a hypothesis?

        “I think most of the physics community would like it to just be something unexpected…”

        Darn tootin’! 😀 (I think “like” undersells it by several orders of magnitude! “Would give a lung for” is probably closer to the mark for many. They’d give both if it would help. 🙂 )

        With our luck, it’ll turn out to be a Trump Bump (i.e. total bullshit).


  4. Thanks for the Coyne link.
    The following is too long as a comment, just a conversation to you personally. So, if it is not suitable as a comment, just delete it.

    The way of confirming a theory is indeed very important, and it consists of two arenas.
    One, empirical knowledge (tests and observations)
    Two, conceptual contemplation (theorizing and speculation)

    From these, there are some derived issues.
    Science: what is science?
    From here, it comes the Popperianism. Thus, I will discuss this issue with 4 steps.
    A, empirical knowledge (test and observation data)
    B, invaliding the nonsense which is unfalsifiable with empirical means
    C, about Popperianism and socially-biased science
    D, about the ‘only game in town’

    Step one: the empirical knowledge is USEFUL but is only a very, very low level tool in the path of searching truths. In fact, it has absolutely ZERO power as a tool to confirm the true (final) truth. I will give three SOLID examples here.

    Example one, the Standard Model (of particle physics) is a phenomenological model; that is, every piece of it is based on empirical data. And, it has passed ALL additional (not the original discovering pathways) throwing monkey wrench challenges thus far (that is, to Today, December 23, 2015). Yet, with this total invincibility from all (any) empirical tests, SM is still viewed as incomplete, totally wrong as a correct (final) theory. How do we know this? Because, it is totally USELESS to perform many genuine TASKs, such as:
    Dark energy and dark mass
    Hierarchy problem
    Calculating nature constant (Cabibbo/Weinberg angles, Alpha, etc.)
    Determining the Neff with an internal framework
    Providing a base (or just a language) for the SM particles
    Giving any hint about the initial condition at or before the Big Bang
    … the list goes on.

    Thus far, the empirical tests and observations are unable to neither supporting the validity of SM nor giving hints for its improvement, and its being a wrong theory is CONFIRMED by its failures of performing those important TASKs listed above.

    Example two: the General Relativity (GR) has also passed (survived) all the monkey wrenches which we can throw at it. If the gravitation-wave is confirmed, it simply passed one more monkey wrench, an icing on the cake. If it is not confirmed (or confirmed as nonexistence), it cannot diminish GR’s great success on so many rounds of victory. Yet, GR is DEFINITELY a wrong theory, playing ZERO role in the correct (final) theory. We know this because:
    Not compatible with the quantum world,
    Playing ZERO role in the very important tasks (open questions) below:
    Dark energy and dark mass
    Hierarchy problem
    Calculating nature constant (Cabibbo/Weinberg angles, Alpha, etc.)
    Determining the Neff with an internal framework
    Giving any hint about the initial condition at or before the Big Bang
    … the list goes on. See,

    Some might say that ‘black hole’ is the god-child of GR. But, ‘black hole’ is a total trash in the correct theory.
    First, it is not directly verified.
    Then, it is totally useless, not being able to help its god-father to perform any tasks listed above. ‘Black hole” plays ZERO role:
    In any viable cosmology model: ‘black hole’ is not even needed in any viable galaxy formation model.
    In any viable dark model model.
    Again, it is the task (not test) criterion cut the ‘black hole’ out to be any useful object in constructing a correct theory.

    Example three: the Higgs nonsense. A new boson (of 125 +/- Gev) was discovered in July 4th 2012, and it was NAMED as Higgs boson. Yet, more than 3 years later, the Higgs mechanism is not verified (empirically). Worse yet, even if Higgs mechanism were right, it would have given ZERO help for making SM a better (complete) theory. Furthermore, Higgs mechanism itself cannot provide a procedure to calculate the mass of the so called Higgs boson. On the other hand, a Vacuum Boson model (VBM) was able to calculate (derive) the mass of a vacuum boson as 125. 46 +/- Gev while the measured mass of new boson is 125.09 +/- 0.24 Gev., see .
    That is, the ‘Empirical test (data)’ itself has no power of judgment, distinguishing the right from the wrong. The greatness of data must be evaluated via other means. The validity of Vacuum Boson is of course supported by its prediction of its mass which was verified with an empirical data. But, the key of its validity comes from its ability to perform all tasks listed above, and I will show the supporting links on this later. In addition to predicting the mass of the vacuum boson mass, the VBM made two other predictions.
    i. The excited state of vacuum energy (vev) is {X (n) = 2(n + 1) X, (X is the ground state)}. So, the X (1) should sit around at 3 X = 738 +/- Gev.
    ii. At ground state vev, the X (1) will not manifest. When the X (1) is reached, the ground state vacuum boson will be greatly reduced (if not disappear altogether) in accordance to a dynamic equation.
    On December 15, 2015, CERN released the RUN 2 (13 Tev) data, and there is a hint about a bump around 750 Gev region. If this new bump is a usual-boson, it should and must show up in the LHC Run-1 from the following reason (rules of thumb).
    R1, the largest particle can be produced in p-p collision is about (1/2) the total energy. So, for 8 Tev, the maximum is 4 Tev particle.
    R2, the production rate increases for the lower mass particle; that is, (1/4) of the maximum should have high enough production rate in a reasonable amount of data (such as 4 to 5 fb-1). That is, anything (with mass) below 1 Tev (under 8 Tev collision energy) should give a hint (bump) with 5 fb-1 and should show up definitely with 20 fb-1 data.
    Thus, with the Run 1 data (about 25 fb-1), we were very confident to announce that there is no SUSY under one (1) Tev. and there is no other USUAL-boson.

    Thus, this Vacuum Boson Model can be confirmed with LHC data in two ways.
    W1, a new bump (around at 750 Gev) is confirmed AND the old bump (125 Gev boson) is reduced.
    W2, when 125 Gev boson showed up, there must not have a 750 Gev bump.
    The W2 is already confirmed with 25 fb-1 data from each lab. In fact, with 4 fb-1 data, LHC Run 1 was already getting a hint for the 125 Gev boson. That is, with 4 fb-1 data for the LHC Run 2, the W1 can be evaluated: the 125 Gev bump should be reduced in relation to the height of the 750 Gev bump.
    Ignoring this fact by CERN is a social issue, nothing to do with the empiricism. Of course, this Vacuum boson vs Higgs nonsense issue will be resolved after more Run 2 data from LHC. Yet, this Vacuum boson issue can still be evaluated now with the ‘TASK’-epistemology (see step three).

    With these three examples, the empiricism is somewhat useful but plays only very low level role in the path of searching for truth, as a squire of the lowest rank. Performing the TASKs is the most powerful tool in the science epistemology.

    Now, step two, any empirical non-falsifiable nonsense can always be invalidated. I am going to use two SOLID examples on this.
    Example A, the KEY point of Multiverse (either as premise or as consequence) is that the Nature constants (of any universe) cannot be calculated (derived) as they are random happenstances. While Multiverse by its definition is beyond any empirical reach, it can be invalidated by showing its premise or consequence is simply nonsense, by showing the following two facts:
    F1, the nature constants of THIS universe can be derived (calculated).
    F2, to show that that calculation is not bubble dependent. See .

    Example B, M-string theory claims that all SM particles are formed as VIBRATIONS of M-string, but it does not have any vibrating-equation (formula) to describe those SM particles, the {string unification}. There is no need of empirical testing their claim as it is a simply failed claim. How can a failed claim be truth? Richard Dawid came up a great idea (the only game in town) to rescue it. The only game in town in the Stone Age is {eat, sleep and sex}, and it is much truer than the M-string nonsense. In fact, the {string unification} is completed, see .
    With these two examples, the empirical non-falsifiable nonsense can easily be invalidated.

    Step 3, about Popperianism
    Coyne wrote: “Pigliucci who, on weak grounds, claim that “Popperism is dead.”
    Massimo Pigliucci was in fact a diehard Popperianism defender, see . I have written the falsifiability issue in detail at: and .
    I am very happy that Pigliucci has changed his position and coming to the side of damning the stupidest idea of mankind, the Popperianism.
    There is NOT a single science advancement which is the RESULT of a (or some) theory being FALSIFIED. Science advances with new theory (replacing the outdate one) being verified. On the other hand, many other theories have been empirically falsified MANY times (such as SUSY) while those devotees are simply moving the goal-post. Popperianism is totally powerless to weed them out. Popperianism is only an ideology for atheism, and it is not only wrong but is totally stupid as a SCIENCE epistemology.
    First, truth (by definition) cannot be falsified. Any theory not representing truth can of course be falsified. A true description of truth can never be falsified.
    Second, the not-testable and the not-falsifiable are two different things. True truth is always testable (with empirical means or with task performing) but not falsifiable (by definition).
    Third, a ‘proposition’ can totally be confirmed with the ‘TASK’-epistemology. For example, the vacuum boson model (VBM) is the result of a unified force equation { F (unified force) = K ħ / (delta T * delta S)} which can perform the following TASKs.
    Part one:
    Calculating the Planck CMB data (Dark energy and dark mass), see
    Calculating nature constant (Cabibbo/Weinberg angles, Alpha, etc.), determining the Neff with an internal framework, see
    Hierarchy problem, see and
    Baryongenesis, see
    String unification, see
    Linking physics to the origin of life, see and
    Gravity, see and

    Part two: ruling out the nonsense (not falsifiable empirically)
    SUSY, see and and and
    Multiverse, see and
    M-string theory, see
    So, Karl Popper’s idea that “a theory must be falsifiable to be scientific” is totally stupid if science is defined as truth searching machine.


    1. Wow, you weren’t kidding when you said it was a long comment 🙂
      There’s a lot here, so these will just be selective responses.

      I totally agree that empirical evidence, by itself, while critically necessary, isn’t sufficient. All observation is theory laden. We can’t escape it. But if we are careful enough in those observations, we can detect when theory isn’t helping us anymore.

      While the physical constants are often a rationale for multiverse theories, my understanding is that it doesn’t drive bubble universe or multiple worlds interpretation multiverses. So demonstrating that those constants aren’t random, while it might remove a philosophical argument for multiverses, wouldn’t itself falsify them. (It’s not clear to me what would. In the case of bubble universes, some people are searching for signs of a collision, which might confirm that theory, but the absence of visible collision wouldn’t falsify it since observable collisions requires profound luck.)

      You said no progress ever happened from a theory being falsified. I would think the orbital precession of Mercury along with the absence of Vulcan, would count. It falsified Newtonian gravitational theory. Yes, per Kuhn, the first response of scientists it to try to patch the existing theory until a better one comes along, but it’s still progress, just not the clean orderly progress some might naively assume. To me, Popper was talking about normative criteria while Kuhn was discussing the sociology of scientists.

      Totally agree with your remarks on only-game-in-town criteria. First, it’s not true anyway, and second it’s too easy for it to just be the only thing we can think of.


Your thoughts?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.