I understand that Popper's falsifiability criterion is meant to demarcate science from pseudoscience. But, is that all one can expect from it?

I mean I do not care about science, but the values it holds, like explanatory power and predictability, many of which can also provided by unfalsifiable statements. For example, take the idea of the multiverse, it holds both explanatory and predictive powers.

So, stating falsifiability as something which distinguishes science from pseudoscience seems shallow and I don't know what to do with it.

  • 4
    Not all good ideas have to be scientific. There are scientifically minded people who also entertain strange and at least currently unfalsifiable ideas. They're usually not zealous about the ideas, they don't get religious about them, they have a sense of intellectual caution and humility about them, but they still take them seriously and may even go as far as to actively believe them.
    – TKoL
    Commented May 15 at 9:44
  • It is not a "rule of method", i.e. some sort of algorithm for sorting out truth from errors. But it can be useful as a maxim, akin to Peirce's Pragmatic maxim or to check fake theories. Commented May 15 at 10:52
  • 6
    Multiverse does not imply anything within our universe, and other universes are supposed to be causally disconnected from ours, which is what makes it unfalsifiable. But if it did imply something about regions of our universe then it would be falsified when such regions do not turn up. The standard of falsification is not absolute certainty, if observable anomalies or variations do not come up long enough - out it goes.
    – Conifold
    Commented May 15 at 10:54
  • physicsforums.com/threads/… Commented May 16 at 10:13
  • 1
    I think a theory that has predictive powers is falsifiable, right? Because it makes predictions which will, at the predicted time, be observed or not; if not, the theory is falsified (a bit of a simplification here, but that's the general idea). There are "multiverse" theories which postulate interactions between the branches that can be observed; Everett's original idea includes that, and is therefore in principle scientific. Theories which don't make predictions are quite pointless though, won't you think? Commented May 16 at 11:50

6 Answers 6


I understand that Popper's falsifiability criterion is meant to demarcate science from pseudoscience. But, is that all one can expect from it?

Falsifiability allows more generally to distinguish science from wishful thinking. A true theory presumably is more useful than a false one. You can always wait for reality to falsify your theory, but it is probably faster and so better to try and falsify it yourself if you can, essentially by submitting it to tests.

I mean I do not care about science, but the values it holds, like explanatory power and predictability, many of which can also provided by unfalsifiable statements. For example, take the idea of the multiverse, it holds both explanatory and predictive powers.

You are contradicting yourself. If a statement is unfalsifiable, it has no predictive powers. If it has predictive powers, it is falsifiable.

Explanatory power has no value in itself. What matters is what sort of explanation it is. God is an explanation and there is absolutely nothing you can do with it because by definition God is not the sort of thing which (if it existed) you could predict what He is going to do. You can only pray and hope for the best! Maybe a much needed exercise in humility?

  • 1
    Wow. Irrelevant and false religion bashing thrown into an otherwise good answer, PLUS a put down of the OP. All religions with Gods involve specific God hypotheses, which are testable. Hence the multiple test cases that religions critics cite, like the Problem of Evil, tests for bad math and physics in scriptures, and internal scripture contradiction tests.
    – Dcleve
    Commented May 16 at 8:18
  • 2
    @Dcleve "All religions with Gods" You are equivocating. "God" is a proper name. It refers to a unique being. Different religions may claim to be the true religion of God, but there is by definition still only one God. As to testing the Scriptures, do you have evidence that Christian churches all recognise that they have definitively been proved wrong? You do wishful thinking well. Commented May 16 at 16:02
  • 1
    @Dcleve "religion bashing thrown into" Words are cheap, isn't it? Do you realise I talked about the idea of God. The word "religion" doesn't even appear in my answer. Discussing the concept of God justifies being charged with religion bashing? - 2. "a put down of the OP" Pointing out a contradiction is a put down? You don't seem to realise that beauty is in the eyes of the beholder. You should try objectivity. Commented May 16 at 16:09
  • 2
    God as a unique being has multiple (conflicting) descriptions among the various monotheistic religions/denominations/cults, so appealing to your own specific idea of this being is itself either wishful thinking or ignorance. I am a Christian who believes that we can predict "what God is going to do" because He has told us about Himself.
    – Jed Schaaf
    Commented May 17 at 4:57
  • 1
    @Dcleve name a single testable hypothesis about a God in a widely recognised world religion. FTAOD "Testable" requires that there be a describable failure mode of the test - a possible situation in which both parties agree that the test is complete and has failed. To put it another way, the test has to have a point at which a reality with a God and a reality without a God behave observably differently.
    – Brondahl
    Commented May 17 at 6:12

Quoting from Section I of the introduction to "Realism and the Aim of Science" by Popper:

The problem of demarcation is to find a criterion that permits us to distinguish between statements that belong to the empirical sciences (theories, hypotheses) and other statements, particularly pseudo-scientific, prescientific, and metaphysical statements; but also mathematical and logical statements.

Nothing about this criterion implies that non-scientific statements are worthless. The book I cited has several sections about good metaphysical theories such as realism. Scientific methodology itself isn't empirically testable as Popper points out in "Logic of Scientific Discovery" Part I Chapter 2. For more of Popper's views on non-scientific theories see "Conjectures and Refutations" Chapters 2, 8 and 11.

  • But he does claim that only those that satisfy his demarcation can be counted as scientific, and even that is too restrictive.
    – Deipatrous
    Commented May 16 at 8:34
  • 2
    It's not too restrictive. The substantive question is "under what circumstances does it make sense to do an experiment or observation?" Popper's answer, which is correct, is that it makes sense when the experimental result could contradict your theory cuz then you might learn something new.
    – alanf
    Commented May 16 at 8:46
  • 1
    The basic error is to mark "scientific" as "good" and "non scientific" as "bad". Popper makes a definition of what "scientific" means. One can see it as shallow. This is disputable, but it is not wrong per se. Commented May 17 at 0:07
  • 1
    It is interesting that he writes "empirical sciences" rather than just "sciences" (+1) Commented May 17 at 8:30
  • It is certainly true that it makes sense to do an experiment or observation if the experimental result could contradict your theory, but it is not true that it makes sense to do an experiment or observation only if the experimental result could contradict your theory.
    – Deipatrous
    Commented May 17 at 12:28

I mean I do not care about science, but the values it holds, like explanatory power and predictability, many of which can also provided by unfalsifiable statements

Not really, that's kinda the point. Like a theory is falsifiable if it makes testable predictions about the real world.

Now predictions are literally Latin for "saying (something) before (it happens)". That is usually what makes theories valuable, they don't just connect the data you already have but allow for making hypotheses of data that you don't already have.

Though doing that comes with the inherent risk of being wrong. That's the problem of induction, just because something like that happened in the past does not certainly mean it's going to happen again in the future. It's not an unfounded assumption, but it's nonetheless an assumption and there is a risk that this theory is wrong.

Now unfalsifiable statements try to avoid that risk. So they either don't make predictions at all, make lots of predictions simultaneously and pick the most convenient as supporting their theory instead ignoring the failure of the rest, making predictions that are untestable idk supernatural causes where you can neither identify the cause, nor observe the action and only see the effect. Or if they do make testable predictions that find ad hoc adjustments and hindsight rationalizations for why this is actually proof of their theory rather than a counter example.

Sidenote: The last one is usually the most tricky for actual sciences, because it hurts to let go of a good theory if it fails and there's some human temptation to just keep it and add an exception or to revise it just slightly to account for that as well. Which is something that is actually done, so "falsifying" is often less of a binary "true/false" and more of a "that's too many exceptions for that rule to be reliable we need to find a better explanation". Not to mention that experimental science often isn't as trivial as the theory makes it sound, so if an experiment fails it could be because of a myriad of reasons not related to the theory and due to the inability of accurate measurements it's also usually not a single number but rather an expectation value and a margin of error in which the result is still accepted as adding confidence to the theory.

Anyway it should be obvious that unfalsifiable statements usually buy that protection from being falsified by surrendering their usefulness. Either they don't make predictions making them pretty useless or making ambiguous predictions which is also useless, are unreliable, which also renders them useless or are concerned with spheres which do not interact with ours, which is also pretty useless.

One prediction stemming from the multiverse hypothesis is the existence of regions within our universe with different physical laws or constants, which could potentially manifest as observable anomalies or variations in cosmological observations.

As conifold has mentioned, that's not what the multiverse hypothesis predicts. universe implies that there is just 1 universe so yes if there would be regions with different physical laws or constants we would expect to see that, either directly or because the interface between the two worlds show some ripple effects of that. In that case you'd have a testable prediction that makes the theory falsifiable.

Or you could argue for the actual multiverse hypothesis that assumes that these regions are disconnected from each other and can't interfere with each other in measurable ways (so multiple universes or one multiverse) in which case you're describing something that can't be tested and doesn't concern our reality, which makes it doubly useless.

  • 1
    This is a good answer except for the last two paragraphs. "Multiverse" can mean different things. ~ Eternal Inflation, it means there are different bubbles, and our universe is one bubble. But there could theoretically be bubble collisions that would leave an imprint on the CMB, or perhaps other evidence like gravitational effects. To date, no evidence has been found, but this particular theory is not unfalsifiable. OP's "multiverse" sounds like the Eternal Inflation version, while this answer seems to be addressing something else, perhaps a "many worlds" theory ...
    – BurnsBA
    Commented May 15 at 19:26
  • But there generally is no neat bijection between hypotheses and prediction. For that reason alone Popperianism can never work.
    – Deipatrous
    Commented May 29 at 19:16

Falsificationism at best provides a demarcation between mature scientific theories and other theories. Even if the multiverse isn't currently falsifiable, it is clearly science - it is part of the search for the best naturalistic explanation of what we see. It is however at an early stage relative to e.g. thermodynamics, but you can bet that those researching it are thinking hard about how it might be falsified. I wouldn't say falsificationism was "shallow" one of the things that is worst about pseudoscience is that its practitioners are typically not self-skeptical and are not looking for ways in which they may be wrong - a hallmark of good science (and basic rational thinking).

Naive falsificationism on the other hand is a bit shallow (IMHO), there is more to science than that and Quine-Duhem suggests we can't unequivocally disprove specific theories either. We also tend not to abandon falsified theories either - they tend to be patched up/modified (at least until the paradigm becomes unsustainable and it needs to be overturned).

The key value of falsificationism is that it stresses (and tests) the importance of self-skepticism. That is a big benefit to science and any form of rational deliberation.

BTW ISTR there is a suggestion that the boundaries of inflationary bubbles in the multiverse could collide and that might leave evidence in the cosmic microwave background, but I can't remember the details. It isn't a given that the multiverse is necessarily unfalsifiable.


Good question. We certainly have to be careful with hypotheses that do not allow for the existence-in-principle of evidence against them, but we cannot do science by throwing out everything that does not satisfy the criterion, at least in its most naive conception.

Things to consider are, e.g. how far down the line of deductions a given falsifiable prediction lies from the postulate, and how much other empirical evidence (or other postulates) get "mixed into the reasoning" before one arrives at that deduction. If every hypothesis lived on an island, so to speak, then the sort of naive Popperism that undergrads learn in Philo 101 might be a solid overarching guiding principle. (Many mature academics don't get much further than thinking falsifiability is the be-all and end-all of empirical science.)

Popper's philosophy, such as it is, owns very much to that of Imre Lakatos; but Popper was just a bit better at making himself famous.


As you are asking for utility, not general "value" or "morals" or anything of that kind, let me point out some very concrete aspects:

So, stating falsifiability as something which distinguishes science from pseudoscience seems shallow and I don't know what to do with it.

Falsifiability puts the onus of investing effort into finding the truth value of a theorem on the creator or proponent of the theorem.

Anybody can state any theory, theorem, worldview, or however you may call it. Unless you also provide some way to falsify it, you leave all the effort of arguing against your theory to your critics. You can then lay back, watch them squirm and waste energy trying to argue against you, and refute anything they say with further theories that you made up on the spot, which do not even need to be true (or falsifiable).

This is one of the major utilities of requiring falsifiability. It removes a possibly huge amount of spurious theories simply based on someone's fancy from earnest discussion.

A second great utility of falsifiability is that it can lead to a cascading effect. Every theory is usually based on some earlier theories, few go back to principles every time. So whenever we newly falsify a theory, we also automatically get rid of all theories based on the falsified one. This is a great time saver in itself.

explanatory power and predictability, many of which can also provided by unfalsifiable statements.

If something has predictability, i.e. if it can make clear predictions (i.e., "if ... then ..." clauses) about a fact we observe in reality, then it is naturally falsifiable. We only need to provide the "if" part and see if the "then" parts follows -> if ever it does not follow, we have falsified the theory.

If something has explanatory power, it either also has predictability; or it is irrelevant for all intents and purposes (for example: any explanation we may eventually have about the big bang either also enlarges our general understanding of the world, and hence gives us predictability; or it may not, i.e. it may be such that it would somehow explain the big bang, but also include nothing whatsoever that can ever occur again due to some internal logic, but this then means that it has no use except to satisfy our curiosity about the past... which may somewhat increase our happiness, but would have no predictability, and hence would be harmless but irrelevant).

So in short, believing that some theory or explanation is not falsifiable but has explanatory power and predictability, is a fallacy in itself.

Final words...

Note that nobody ever said that science is the only relevant thing in life. In fact, every human works with non-scientific beliefs and belief systems on a daily basis, and most of the time this works very well. It saves a lot of time, it makes quick decision-making possible in case of restricted or incomplete information; and as long as you do not try to influence other people negatively based on made-up theories or beliefs, there is little harm you can do with it (except to yourself, which as an adult is your prerogative).

  • I am asrounded to see that so many people really seem to think that Popper's criterion actually does provide the demarcation that separates science from nonscience. Would that it were so! I also wonder how many of these Popperians have even one scientific breakthrough to their name...
    – Deipatrous
    Commented May 29 at 19:13
  • @Deipatrous, I mean, it's not like science is there and you need to find a demarcation. It's more like F. is a relatively plausible definition, which ties in well with practicalities. I.e., it gives working scientists a clear method of work, they know when they are "finished", right? And OP does not ask whether F. is right or not (they already seem to have their mind pretty much made up that F. is "shallow" at the very least). This answer is about practical implications or benefits of defining science as that which is demarked by falsifiability.
    – AnoE
    Commented May 31 at 8:41
  • But it does not tie in with the practicalities of science at all. It is neither descriptive, nor normative i.e. it could not even work in an ideal world. I wish the popperian criterion did mark the boundary between pseudoscience (nonsense, mysticism and so on) because that would be an easy end to endless debates, but it just isn't so.
    – Deipatrous
    Commented Jun 1 at 15:23
  • Hm, enforcing that a theory contains a recipe to (in)validate it seems practical to me? Maybe we're talking about different things. Is there a better alternative, @Deipatrous?
    – AnoE
    Commented Jun 3 at 8:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .