{X} "constitutes a falsification of" {Y} - (good) science?

[Disinfo Agent]

For example, this thread, regardless of it's "intent", has made a powerful (and so far unsuccessfully challenged) case that "naive falsification" is not the process whereby science advances.

Popper's major work on science is called in English The Logic of Scientific Discovery. As the name suggests, it's about the logical justification of the scientific method. What makes science so successful and persuasive? What allows us to assign a greater weight to science than to, say, astrology or divination or gut feelings? How do we know that ID proponents are wrong when they claim that "science is just a belief system as valid as any other religion"? How do we separate the wheat from the chaff? These were the kind of questions he addressed.

It was the problem of drawing a line (as well as this can be done) between the statements, or systems of statements, of the empirical sciences, and all other statements — whether they are of a religious or of a metaphysical character, or simply pseudo-scientific. Years later — it must have been in 1928 or 1929 — I called this first problem of mine the "problem of demarcation." The criterion of falsifiability is a solution to this problem of demarcation, for it says that statements or systems of statements, in order to be ranked as scientific, must be capable of conflicting with possible, or conceivable, observations.

That scientific theories don't get falsified every day of the week, or that falsification is not a childish black and white choice, is hardly surprising, or a contradiction of his ideas.

[Ken G]

My example was mostly to point out that it is sometimes not clear what is being falsified.

Now I see what you are driving at, sometimes there are multiple possibilities for what was "falsified", and we don't always get to know which one is the real "culprit". That is certainly an important point, as there are many great historical examples of thinking one thing was falsified when in fact it was something else (the Greeks thought the absence of stellar parallax falsified the idea that the Earth was in motion, whereas what it really falsified was the idea that stars are just outside the bounds of our solar system).

In relation to the CDM business, one might say that we have falsified the union of Newtonian gravity with no dark matter, but we don't know which of those possibilities has actually been falsified. dgruss23 seems to argue that the latter has gained some additional support from failures of other experiments, which tends to support the idea the former is the culprit of what has been falsified. Others might point to the bullet cluster and other such observations to point to the latter as the source of the falsification. So yes, there can certainly be an important debate about what has been falsified that can go on for a long time, even millennia.

[Ken G]

Popper's major work on science is called in English The Logic of Scientific Discovery. As the name suggests, it's about the logical justification of the scientific method.

Well that's his problem right there-- science uses logic, but is not justified by logic. The proof is in the pudding, when it comes to science-- it is justified by how it accomplishes its purposes. (This is starting to remind me of our debate about the differences between science and pure mathematics-- the former uses nature as the adjudicator of success, not logic. Logic is the tool we use to build science, not what we use to judge science.)

How do we know that ID proponents are wrong when they claim that "science is just a belief system as valid as any other religion"?

It has nothing to do with the "logic" of science, I can tell you that! We know they are wrong when they say that because that is not what science is, they may as well be saying that an elephant is a kind of fish. The key difference between science and a "dead weight" theory like ID is that ID is not useful for anything, except giving a warm fuzzy feeling that we are being looked after somehow. Science must always do more than convey a warm fuzzy feeling, it's successes are testable. That's part of its purpose, not it's "logic".

That scientific theories don't get falsified every day of the week, or that falsification is not a childish black and white choice, is hardly surprising, or a contradiction of his ideas.

It was you who brought Popper into this discussion-- the thread is about "naive falsification", so is exactly about black and white choices.

[Disinfo Agent]

[...] science uses logic, but is not justified by logic. The proof is in the pudding, when it comes to science-- it is justified by how it accomplishes its purposes.

Nonsense. Any run-of-the-mill religion "accomplishes its purposes", too. Does that make it science?

It was you who brought Popper into this discussion-- the thread is about "naive falsification", so is exactly about black and white choices.

Don't be naive.

[HenrikOlsen]

Don't be naive.

Careful.

[dgruss23]

In relation to the CDM business, one might say that we have falsified the union of Newtonian gravity with no dark matter, but we don't know which of those possibilities has actually been falsified. dgruss23 seems to argue that the latter has gained some additional support from failures of other experiments, which tends to support the idea the former is the culprit of what has been falsified. Others might point to the bullet cluster and other such observations to point to the latter as the source of the falsification. So yes, there can certainly be an important debate about what has been falsified that can go on for a long time, even millennia.

It is always remarkable to me on this board how the real meat of what a person posts sometimes fails to get chewed while a bit of extraneous grizzle is worked to death. (and I mean that in reference to this thread not any one person's post). You guy's have been chewing the heck out of my use of the word "falsification" yet the observations and papers I linked to in the other thread are getting far less attention in that thread.

KenG, I certainly don't have a problem with what you said earlier:

Newton's gravity, on the other hand, has been "falsified", yet it is not dead weight, because it continues to be useful and we know exactly when we can use it and when we cannot. In short, it accomplishes a scientific purpose, while those others do not. For the time being, CDM also accomplishes its purpose. That may not always be so, but at the moment dgruss23's objections are premature (not completely invalid, just premature).

I think this is an important point to consider. First let me expand the quote from the OP another sentence or two:

And that is the issue with all of these cases I've noted and it constitutes a falsification of the expectations for CDM particles - and any collisionless DM particles. If DM is collisionless it should not be coupled to the distribution of luminous matter. On the scale of individual galaxies CDM is a failure.

So I want to be clear that the emphasis - and anyone who has read the other thread through will know this - that I'm outlining problems for CDM on galaxy scales. My comment was not meant to be a sweeping statement that CDM is falsified across all regimes of study.

But now to the topic of this thread. I actually like KenG's reference to the "usefulness" of a theory in the above quote much better than "falsification". For example, introductory chemistry classes instruct students on the combined gas law and the ideal gas law even though these gas laws are "falsified" due to the fact that real gases experience particle attractions that cause them to deviate from the ideal gas law. A better equation is the van der Waals equation which is a modification of the ideal gas law.

However, the ideal gas law is still useful because it is a good approximation (at least for pedagogical purposes). Newton's laws fall under this category - they still have practical application because we can make accurate enough predictions with them for many purposes.

CDM also is still useful on the scale of galaxy clusters and superclusters and where cosmology is concerned. CDM simulations seem to reproduce the large scale structure. CDM helps to fulfull the needs of a flat inflationary universe. CDM allows cosmologists to make predictions that can then be observationally tested. All this works on galaxy cluster scales - and there are successes on that scale that justify the usefulness of CDM.

But my point in the other thread is that on galaxy scales - CDM has met with one failure after another when the expectations are compared with observations. The expectations for CDM particles are not verified on galaxy scales. The other thread is where we're discussing that evidence specifically.

However, given those failures - and speaking to the topic of this thread - CDM is not very useful for understanding the observations of the dynamics of individual galaxies. You can make predictions with CDM for galaxy dynamics, but those predictions are repeatedly contradicted by observational results. And that is why I used the word "falsification". What use is a hypothesis which has predictions that are repeatedly shown to be in conflict with observations?

Now if someone thinks it is premature to say CDM cannot be made to fit with those observations being discussed on the other thread, that is fine. However, the motivation for continuing to use CDM on galaxy scales is not because of any galaxy scale successes, but because of the compatibility of CDM with galaxy cluster observations and cosmological theory.

My purpose in the other thread was to point this out. CDM is one of the key components of cosmological theory. And like it or not - galaxy scales is part of the foundation upon which the continued usefulness of CDM rests.

[dgruss23]

Very interesting... and here I thought dgruss despised Popper.

Hmmm ... on what basis?

[Disinfo Agent]

None whatsoever, it turns out. I was relying on my memory, but a couple of searches in the forum have convinced me that my recollection could not have been more wrong. I guess I must have mixed up your criticism of Occam's Razor with criticism of falsificationism.

My apologies.

[Disinfo Agent]

[...] science uses logic, but is not justified by logic. The proof is in the pudding, when it comes to science-- it is justified by how it accomplishes its purposes.

Nonsense. Any run-of-the-mill religion "accomplishes its purposes", too. Does that make it science?

Rereading what Ken G wrote, perhaps I misinterpreted him here. The word "how" has just jumped at me.

Ken, when you write that science 'is justified by how it accomplishes its purposes', do you mean that the method used by science is what justifies it? (I initially understood it as meaning that science is justified by the fact that it accomplishes its purposes.)

[Ken G]

Nonsense. Any run-of-the-mill religion "accomplishes its purposes", too. Does that make it science?

The point I was making is that science is judged by the accomplishment of scientific purposes, not just any purpose. I thought I was quite clear that simply providing a feeling of comfort was not sufficient to be considered a purpose of science.

[DrRocket]

It's a general question: what - in your opinion - is the role of 'falsification' in modern science?

It seems to me that as a practical matter that "falsification" is used in two ways.

Practicing scientists and other rational people recognize that a theory that fails to conform to what is accurately and appropriately observed is at least somewhat flawed. The response to that recognition is to attempt to devise a more accurate theory. It does not mean that a theory, such as Newtonian mechanics, that is known to not be correct in all circumstances is abruptly tossed out and forgotten. However, it does mean that a new or proposed theory that fails to predict phenomena in a regime in which it was intended to provide accurate predictions ought to be abandoned or substantially revised. None of the current physical theories is "true" in the mathematical sense and all have been falsified in the same sense that Newtonian mechanics has been falsified. We know that quantum theory and general relativity are incompatible. But neither is about to be thrown on the rubbish pile.

Other people, on the fringe, seem to interepret the ideal of falsification to suit their personal and immediate agendas. Because current theories are inadequate to describe all that is observed (for instance the notions of dark energy and dark matter are used to in an attempt to describe a state of current ignorance) they feel justified in discarding most, if not all, of mainstream physics and inserting their own ad hoc notions of how nature works. This is clearly an inappropriate and irrational application of "falsification." It is perhaps traceable to the use of falsification in mathematics, in which acceptance of a theorem is quite binary. A theorem has either been proved or not proved and any counter-example is enough to declare a hypothesis to be false. But mathematics is not science.

[Ken G]

Ken, when you write that science 'is justified by how it accomplishes its purposes', do you mean that the method used by science is what justifies it? (I initially understood it as meaning that science is justified by the fact that it accomplishes its purposes.)

Your initial reaction was the correct one-- the latter reading would be more along the lines of what it sounded like you were saying Popper was saying. We cannot judge science by its techniques, because we define science by its techniques, and you have to judge something differently than you define it or it is a tautology. We set goals for science, and then judge how it measures up-- and it has different goals than religion. The goals of science are to unify, simplify, understand, and predict, as quantitatively as necessary to accomplish some useful purpose, the outcomes of experiments and observations that are objective and repeatable. The experiments and observations are the interface with nature that is the ultimate judge and jury of scientific theories. That's a more sophisticated way to say "falsification", and I really don't see what is so profound or difficult about it that it requires the machinery of philosophy to ascertain. Nor does it sound at all like religion.

[Ken G]

A theorem has either been proved or not proved and any counter-example is enough to declare a hypothesis to be false. But mathematics is not science.

That point is also quite relevant to the "sidebar" I'm having with Disinfo Agent.

[Ken G]

Now if someone thinks it is premature to say CDM cannot be made to fit with those observations being discussed on the other thread, that is fine. However, the motivation for continuing to use CDM on galaxy scales is not because of any galaxy scale successes, but because of the compatibility of CDM with galaxy cluster observations and cosmological theory.

I confess I haven't read those links you refer to and I haven't been participating in that other thread, but I think there is some danger of "co-opting" this thread. I realize that this thread is getting a lot of comments and is reaching conclusions that you feel are giving a skewed perspective of what that other thread is actually all about. At the risk of participating in that detour, my level of understanding of CDM in galaxies is simplistic, but the "standard canon" on that topic is that any mass distribution that falls like 1/r^2 will explain a constant-velocity rotation curve. A 1/r^2 falloff is expected for an isothermal gas in pressure equilibrium with its own gravity. So that sounds like too unifying an explanation to discount, even if it is at a superficial level, and makes CDM seem like it is not "dead weight" even on galactic scales.

[Nereid]

It is always remarkable to me on this board how the real meat of what a person posts sometimes fails to get chewed while a bit of extraneous grizzle is worked to death. (and I mean that in reference to this thread not any one person's post). You guy's have been chewing the heck out of my use of the word "falsification" yet the observations and papers I linked to in the other thread are getting far less attention in that thread.

[...]

Curious, isn't it?

There's a part of this post of dgruss23's that I think is very relevant to this thread, but I'll bring it up in a separate post.

[Nereid]

[...]

KenG, I certainly don't have a problem with what you said earlier:

Newton's gravity, on the other hand, has been "falsified", yet it is not dead weight, because it continues to be useful and we know exactly when we can use it and when we cannot. In short, it accomplishes a scientific purpose, while those others do not. For the time being, CDM also accomplishes its purpose. That may not always be so, but at the moment dgruss23's objections are premature (not completely invalid, just premature).

[...]

I actually like KenG's reference to the "usefulness" of a theory in the above quote much better than "falsification". For example, introductory chemistry classes instruct students on the combined gas law and the ideal gas law even though these gas laws are "falsified" due to the fact that real gases experience particle attractions that cause them to deviate from the ideal gas law. A better equation is the van der Waals equation which is a modification of the ideal gas law.

However, the ideal gas law is still useful because it is a good approximation (at least for pedagogical purposes). Newton's laws fall under this category - they still have practical application because we can make accurate enough predictions with them for many purposes.

CDM also is still useful on the scale of galaxy clusters and superclusters and where cosmology is concerned. CDM simulations seem to reproduce the large scale structure. CDM helps to fulfull the needs of a flat inflationary universe. CDM allows cosmologists to make predictions that can then be observationally tested. All this works on galaxy cluster scales - and there are successes on that scale that justify the usefulness of CDM.

But my point in the other thread is that on galaxy scales - CDM has met with one failure after another when the expectations are compared with observations. The expectations for CDM particles are not verified on galaxy scales. The other thread is where we're discussing that evidence specifically.

However, given those failures - and speaking to the topic of this thread - CDM is not very useful for understanding the observations of the dynamics of individual galaxies. You can make predictions with CDM for galaxy dynamics, but those predictions are repeatedly contradicted by observational results. And that is why I used the word "falsification". What use is a hypothesis which has predictions that are repeatedly shown to be in conflict with observations?

Now if someone thinks it is premature to say CDM cannot be made to fit with those observations being discussed on the other thread, that is fine. However, the motivation for continuing to use CDM on galaxy scales is not because of any galaxy scale successes, but because of the compatibility of CDM with galaxy cluster observations and cosmological theory.

[...]

(colour added; bold in original)

Before I continue, I want to make it clear that I have snipped parts of dgruss23's post to highlight an aspect of "falsify" that I am interested to discuss in this thread; let's keep discussion of CDM etc to the other (Q&A) thread.

From the OP:

Where galaxy scales are concerned it is not the "best" - MOND is actually better.

[...]

And in the case of collisionless CDM and MOND, how helpful is it that MOND was falsified before it was even written down^?

[...]

^ MOND is inconsistent with special relativity (SR), and SR has passed an enormous range of experimental and observational tests.

I think this rather neatly illustrates an inherent shortcoming of "falsify" (keeping in mind the distinction I made earlier between "falsifiable" and "falsified"):

* as a hypothesis, CDM on galaxy scales may (or may not) be consistent with (lots of) good observations

* as a theory (or part of a theory), CDM may be very useful, as a device for generating lots of really good new research ideas

* as a hypothesis, MOND is surely falsified because it is in intolerable conflict with SR

* as a phenomenological heuristic, MOND may be fantastic (what was it called? Milgrom's fitting formula?), and may spur lots of really good new research ideas.

If the yardstick is "falsify", then maybe we need to dump CDM on galaxy scales, and we certainly need to dump MOND. But how useful is that?

So, to paraphrase what I wrote in post#13, from what's in this thread (including dgruss23's post quoted here), it seems that "X (observations) constitutes a falsification of (theory/idea) Y" is neither an accurate description of what happens in science nor a useful guide to how to actually do science.

[Ari Jokimaki]

So, to paraphrase what I wrote in post#13, from what's in this thread (including dgruss23's post quoted here), it seems that "X (observations) constitutes a falsification of (theory/idea) Y" is neither an accurate description of what happens in science nor a useful guide to how to actually do science.

It seems to me that generally falsification should be applied to detailed theories only. I don't see much point in trying to falsify "ideas", because I doubt that "ideas" are usually falsifiable (but sure there might be some ideas that are detailed enough so that they too can be falsifiable). When I think of "idea", I think that we are in more of a philosophical domain than in physical. To say that an idea or concept is falsified because it seems to go against some current observations doesn't have much meaning if the idea or concept is not specified accurately enough to be genuinely falsifiable. With that in mind, I quote couple of things from myself (in another forum about a year and half ago):

I don't see the point in rejecting the concepts which might not be falsifiable. Especially in cosmology there is not much we can actually falsify. I think we have to be very careful about falsifiability when we discuss philosophical concepts. Some models of universe might contain philosophical concepts that are not falsifiable, but usually those models also contain physical concepts that are falsifiable, so generally you can falsify a model even if it would contain some unfalsifiable concepts. For example, expansion of space is not falsifiable concept; even if you would show that space doesn't expand somewhere, one could argue that it expands somewhere else, just not there where it was measured.

and:

One additional point about falsifiability: when we say that something is not falsifiable, we really mean that we can't falsify that something with our current level of knowledge. But when our level of knowledge increases, so does our ability to falsify things. So, if something doesn't seem falsifiable currently, it might be falsifiable in future, therefore it makes no sense to reject concepts based on falsifiability alone.

[Ken G]

To continue a bit on this "falsifiability" issue (rather than the core "falsified" issue of the thread), I see the former as a crucial element of science, it merely means we need to distinguish what we might now falsify versus what is falsifiable in principle. They are both science, but two different "flavors" of science, with different goals. In terms of the unfalsifiability of "expansion of space", I think you have reached the wrong conclusion: you've concluded it is science anyway, but I would say that in fact you have put your finger on exactly why it is not real science to say that "space is expanding", that is merely one picture you can use to describe the real science. The real science is that distances are increasing, for whatever reason you like to picture that and is supported within the general theory that works (shrinking rulers works fine too, for example). I would call that a "pedagogy", a way to teach a theory that is not the same as the theory because it is not testable in the same way. The way you test a pedagogy is called a "final exam", the way you test a theory is called an "experiment".

[Ken G]

So, to paraphrase what I wrote in post#13, from what's in this thread (including dgruss23's post quoted here), it seems that "X (observations) constitutes a falsification of (theory/idea) Y" is neither an accurate description of what happens in science nor a useful guide to how to actually do science.

Yes, to me a far better word for what observations do is "constrain" theories, not "falsify" them. They quantify the usefulness of a theory in objective and repeatable ways, and that's the true core of science. Why would we put all our attention on "falsification" when what we really need to recognize in science is the need for a dialog with nature. Making nature the ultimate authority is the core genius of science-- what our brains do to organize and unify that is just what we do with what nature is "telling us".

[Nereid]

So, to paraphrase what I wrote in post#13, from what's in this thread (including dgruss23's post quoted here), it seems that "X (observations) constitutes a falsification of (theory/idea) Y" is neither an accurate description of what happens in science nor a useful guide to how to actually do science.

It seems to me that generally falsification should be applied to detailed theories only. I don't see much point in trying to falsify "ideas", because I doubt that "ideas" are usually falsifiable (but sure there might be some ideas that are detailed enough so that they too can be falsifiable). When I think of "idea", I think that we are in more of a philosophical domain than in physical. To say that an idea or concept is falsified because it seems to go against some current observations doesn't have much meaning if the idea or concept is not specified accurately enough to be genuinely falsifiable. With that in mind, I quote couple of things from myself (in another forum about a year and half ago):

I don't see the point in rejecting the concepts which might not be falsifiable. Especially in cosmology there is not much we can actually falsify. I think we have to be very careful about falsifiability when we discuss philosophical concepts. Some models of universe might contain philosophical concepts that are not falsifiable, but usually those models also contain physical concepts that are falsifiable, so generally you can falsify a model even if it would contain some unfalsifiable concepts. For example, expansion of space is not falsifiable concept; even if you would show that space doesn't expand somewhere, one could argue that it expands somewhere else, just not there where it was measured.

This points to another aspect that we have touched on but not explored much in this thread: scope.

Do hypotheses get to be falsified? theories? ideas? concepts? models? laws?

An experiment or observation may be quite clear-cut: Mercury's perihelion moves in this way (with stated uncertainties/error bars), the atomic clocks aboard the planes differed by this many seconds when they got back home, the A, B, C and D images of the distant quasar varied over time in this way, and so on. But how do these clear-cut results relate to GR (for example), or CDM?

I think a good example is the Pioneer anomaly: analysis of the tracking data shows an unmodeled acceleration of {value and direction; uncertainties}.

So something is a bit off, or odd ... but what?

Enter hypotheses, which are very specific, quite narrow in scope, and as fully quantitative as we can make them - they provide clear guidance, or pointers, on what to do on Monday when we get to the lab, because they are eminently testable (albeit sometimes more in principle than in practice).

As a general rule, the things to be tested first are what's in textbooks - on Monday you build models using standard physics.

Suppose you've done that, written papers, chatted over coffee (or something else) at meetings, etc, etc, etc, and the day comes when models built on standard physics have all been ruled out; what then? Time to sketch some ideas for new physics? which will be severely tested in the crucible of your colleagues sharp minds (and, often, tongues) ... modify, re-state, get the idea torn to ribbons again, repeat. Maybe one day a theory emerges ... what then?

Can we say that "neutrinos" were such a theory, from the 1930s? or is that too grand a word?

Certainly neutrino flavours/oscillations were a theory long before 2001!

Is MOND a theory?

Is CDM a theory (non-baryonic CDM, whether collisionless or not)?

Are 'gap fillers', or 'place holders' in some kind of limbo, neither hypotheses nor theories? extremely useful but not really falsifiable? They are undoubtedly convenient as shorthands, if nothing else!

and:

One additional point about falsifiability: when we say that something is not falsifiable, we really mean that we can't falsify that something with our current level of knowledge. But when our level of knowledge increases, so does our ability to falsify things. So, if something doesn't seem falsifiable currently, it might be falsifiable in future, therefore it makes no sense to reject concepts based on falsifiability alone.

Yep ... and that's one reason why I try to be careful to always add "in principle"!

[hhEb09'1]

One additional point about falsifiability: when we say that something is not falsifiable, we really mean that we can't falsify that something with our current level of knowledge. But when our level of knowledge increases, so does our ability to falsify things. So, if something doesn't seem falsifiable currently, it might be falsifiable in future, therefore it makes no sense to reject concepts based on falsifiability alone.

That's not really how I'd use it.

The things I'd classify as not falsifiable are invisible elf types, or religious beliefs maybe. And I don't reject those sort of things, out of hand. They just get put in the nonscience bucket.

PS: Smiley-free posting for the past **015** posts

[dgruss23]

I feel like a rather straightforward concept about the process of science is getting lost in all this discussion. And that would be the role that observations play in determining the fate of a hypothesis/model/theory/principle/law ...

In the context of what this thread was broken off from one starts to wonder if perhaps some people posting on this thread actually think that observations cannot rule out a hypothesis/model/theory/principle/law. Now I know that is not actually the case, but if this is supposed to be a discussion of the validity of what I said on the other thread, then the context is getting completely lost.

When a hypothesis is proposed, it's usefulness as an idea comes not only from what it predicts should happen, but also from what it predicts should not happen. When observations are made to test the hypothesis, the scientist must ask if within the uncertainty of the observations do the observations:

1. Confirm the expectations of the hypothesis (hypothesis X says Y should happen and Z should not happen. Observations show that Y happens and Z did not happen.)
2. Fail to allow a determination of the expectations of the hypothesis (Y might have happened but we cannot be sure and/or Z might not have happened but we cannot be sure).
3. Contradict the expectations of the hypothesis (Y did not happen, and/or phenomenon Z that should not happen actually did happen).

If #3 is what happens then you have a hypothesis that is in conflict with observations. If #3 is confirmed multiple times and there are no addendums that can be made to the hypothesis to turn #3 from a contradiction into a compatible result then it must be concluded that the hypothesis is not useful, incorrect, falsified, fails to match the real universe ... pick what you think is the most appropriate term.

In the other thread I discussed observational results that I argued fall under category #3 above. I had rather expected that we might be able to have some discussion/debate as to whether or not others agree/disagree that those observations are in fact an example of category #3. Instead, the discussion has turned down the road of suggesting that a scientific hypothesis which is in direct conflict with observations, fails to make predictions that observations confirm, and cannot explain certain observations should nonetheless not be considered "falsified", or incorrect, or no longer useful.

Moti Ben-Ari discusses falsification in his book "Just A Theory" which is a defense of science against creationists and others that attack science itself. He says the following things in his section on "Falsification as a Methodology":

Scientists propose a theory and then perform experiments to check predictions of the theory. If an experiment confirms the predictions, there is no incentive to change the theory. If, however, an experiment refutes the predictions of a theory, the theory is falsified, and scientists must search for a new theory not refuted by the experiment (nor by any previous experiments that had confirmed previous theories).

...

While it is true that scientists do engage in attempts to falsify existing theories, as a methodology falsification simply does not describe what actually happens in the process of science. First, scientists are primarily motivated by the ambition to propose and confirm theories that they themselves believe to be true, not falsify theories suggested by others.

...

A second difficulty with falsification as a methodology is that the validity of a theory is not judged simply by checking experimental results against tthe theory, but by a complex process of mutual feedback within the community of scientists. The development of a scientific theory requires intense mental effort, so that it comes as no surprise that a scientist may be so emotionally attached to a theory that occasional or even frequent falsifications will not shake him from his belief in the correctness of the theory.

Notice that Ben-Ari confirms that falsification is not how science works in practice (as suggested by Nereid and others) - but the reasons given are not arguments that the ideal of falsification is incorrect. Rather Ben-Ari points out that scientists are people too with motivations beyond the idealistic notion of falsification.

It is my hope that perhaps some will engage in the discussion I expected to have as highlighted in green above.

[Ken G]

I think the issue there is that most of us would rather discuss science in general, with applications to all aspects of it, rather than some specific application like CDM. Quite frankly, there are no aspects of CDM that give me a personal issue, though it certainly could be due to lack of knowledge of the observations on galactic scales that you refer to. However, I do have great issues with the way the process of science is often framed in various places.

The account above by Ben-Ari does nothing to alleviate my disquiet. He stresses two personal aspects of doing science, both centered on a kind of mental comfort or allegiance of some kind, as experienced by human scientists in the act of doing science. While those are both valid human limitations at doing good science that are worth pointing out, it is ill-advised to discuss them as if they defined the process, or represented it at its best. Instead, they are more like pitfalls to watch out for in the act of doing good science.

So they leave unanswered the real question of what defines good science, and why has it been so successful to this point. I don't think either of the red highlights even touch on those issues. But one thing is abundantly clear-- a central and dominant role for the unbiased and openminded interpretation of the "voice of nature" is the crucial underlying theme in all good science. Whether that is framed in terms of "falsification" or "confirmation" or "mental effort" is all fairly irrelevant, compared to simply asking : does the primary motivation for some model or theory come from the way it helps us unify and comprehend in a useful way how things actually are, rather than how we'd like to imagine them or what will make a better press conference or what generates less cognitive dissonance to think about? There is no "magic bullet", no failsafe recipe, for answering that question, we simply have to step back and be as objective and clear-minded as possible and do our best to answer it. But the core question involves usefulness of a theory, not correctness (which is too black-and-white of a notion), and that is the main flaw in "naive falsification".

[Disinfo Agent]

The point I was making is that science is judged by the accomplishment of scientific purposes, not just any purpose.

[...]

Your initial reaction was the correct one-- the latter reading would be more along the lines of what it sounded like you were saying Popper was saying. We cannot judge science by its techniques, because we define science by its techniques, and you have to judge something differently than you define it or it is a tautology. We set goals for science, and then judge how it measures up-- and it has different goals than religion.

Here's a slightly hypothetical but not altogether implausible situation:

a) Science says that humans are causing global warming, and we'll be in trouble soon if we don't do something to counter it.

b) Some religion says that global warming is actually a test to mankind's faith, and that we'll be saved by God in the end so long as we don't try to intervene, but we'll be punished with worse things it we try to do anything about it.

c) A doomsday cult agrees that global warming will be the demise of mankind, but says we should embrace it, because it's a sign that the Rapture is at hand, and any interference will be pointless anyway.

How do you choose between these three propositions? Which one will you believe, and why?

[Ken G]

There is no "magic bullet" to choose between those, it all depends on your goals. People who think that there is some "logical way" to prove that science gives you the "right answer" are sadly mistaken-- it is illogical to imagine such a possibility could flow from logic alone. Logic never gives you answers, it is a technique for manipulating axioms such that all the answers are already determined by the axioms and you are merely trying to ascertain the connections you don't recognize without the assistance of a logical procedure. Thus the goal of logic is to help you understand your axioms, not to know if they were the "right axioms". Insofar as your list above is really a list of axioms, not a list of results, there is no way to decide "which is right" unless such a test is expressly built in as one of your axioms. That's what science does, and what religion does not do, and that is by far their most important difference. Both lead to perceived benefits-- one leads to benefits that are by axiom objectively and repeatably testable, the other to benefits that may only be tested on a personal basis, or not at all.

[DrRocket]

There is no "magic bullet" to choose between those, it all depends on your goals. People who think that there is some "logical way" to prove that science gives you the "right answer" are sadly mistaken-- it is illogical to imagine such a possibility could flow from logic alone. ....

I think there is an aspect to your assertion that has thus far gone unsaid. There does seem to be some popular belief that science can give you the "right answer", but that reflects a terrible popular misconception. And that misconceptioin is that there actually is some absolute "right answer". This comes back to the earlier point that science is not mathematics.

In mathematics one can and does work with something approaching "absolute truth". All theorems are carefully stated and are either proved or not and can be, via counterexample, shown to be wrong. But all scientific theories, at least all that exist at this time, are known to be only approximationss to the reality of nature. Unlike approximations that occur in mathematics, approximation in science are not always precisely quantified a priori. We know that Newtonian mechanics is less accurate than general relativity. Yet we use Newtonian mechanics in a myriad of situations, secure in the knowledge that it is good enough, without a mathematically pure definitioin of what "good enough" is. We know that Newtonian mechanics is "wrong" because we know that general relativity is better. But we also know that general relativity is probably wrong, because it is incompatible with quantum mechanics, and we don't have anything better.

The point here is that falsifiability in science is not an absolute, at least as a practical matter and as actually practiced. Some things in science are more false than others. Science progresses as a series of successive approximations, and that fact is lost on many in the general populace.

[Ken G]

I think there is an aspect to your assertion that has thus far gone unsaid. There does seem to be some popular belief that science can give you the "right answer", but that reflects a terrible popular misconception. And that misconceptioin is that there actually is some absolute "right answer". This comes back to the earlier point that science is not mathematics.

We are in complete agreement. What's more, this misconception about science is very much a "double-edged sword", because if there is an absolute "right answer", then all the others are wrong. That reminds me of how people think about their own religion! One might think that much of the authority of science stems from thinking this way, so relaxing it weakens the relevance of science, but I claim that is exactly the opposite of the truth. I see much of the rebelliousness we see against science stemming from the idea that there is a "right answer", such that people turn to other means of obtaining that answer if they like it better and are forced to choose. Rather, if one understands what science really is, one sees that we are not choosing what to believe, we are choosing our objectives for forming a belief in the first place. The rest follows with little cognitive disquiet.

The point here is that falsifiability in science is not an absolute, at least as a practical matter and as actually practiced. Some things in science are more false than others. Science progresses as a series of successive approximations, and that fact is lost on many in the general populace.

Yes, indeed I've heard it said that the actual meaning of an "exact solution" in science would be "tell me how much error you will tolerate in the solution, no matter how small, and I can give you an answer that satisfies it." Note that such a prescription cannot even get started without first being told the desired accuracy target. So a theory that is "falsified" within one accuracy target might be "confirmed" within some looser target (history is replete with that), generating a kind of "onion-skin" structure to scientific knowledge.

But lest we get too hung up on the issue of accuracy, note that science is about more than just the error in the prediction. It is also about organizing information, and suggesting new avenues for study. A good theory does this as well, and if it does it well, larger inaccuracies become tolerable. We also heard that sometimes it's not obvious what aspect of the theory is causing the problem, perhaps some idealization was made in a theory that is otherwise quite good.

As long as a theory is generating the right questions to lead to advancement and progress, it is doing its job-- and that may be more the issue that comes up in the context of CDM. In fairness, however, it would also seem true that slavish attachment to a particular way of thinking even after it has passed its usefulness is a pitfall to avoid as well (personally I don't see that for CDM, even on galactic scales, but perhaps I'm not privy to enough information on the topic).

[Warren Platts]

I see much of the rebelliousness we see against science stemming from the idea that there is a "right answer", such that people turn to other means of obtaining that answer if they like it better and are forced to choose.

The word 'truth' serves a useful function in the language--even within nonscientific contexts. It is not something to be given up lightly. And different answers definitely have social consequences. Theory is underdetermined by the evidence. (That much we can be sure of.)

And so why shouldn't people "turn to other means" to obtain the truth? The rebelliousness you mention is mainly because of the perceived ethical consequences that ensue--not from science per se--but from the perceived ethical consequences of scientifically inspired metaphysical systems such as atheism, naturalism, and materialism.

And since theory is underdetermined by the evidence, given the choice, should not one choose the more ethical alternative???

Naturalism says that what is real is just what the scientists say is real; and if there is one essential mark to science, it is the link between theory and observation. Obviously, that link does not merely consist of naive falsificationism. Nevertheless, naturalism looks only to science to say what is and how we know what is. Ultimately, this is a pragmatic stance. We choose naturalism not because there is a proof somewhere that naturalism is true; no, we choose naturalism because focusing on the link between theory and the stimulation of nerve endings seems to work better than the stimulation of the internal brain cells, as when one claims to "feel" that Jesus has just entered one's soul or whatever.

Both are kinds of experience, however.

So what makes science different from other epistemologies is not merely the connection between theory and experience--no, it is the connection between theory and a certain kind of experience having to do with nerve endings at the interface between one's body and the rest of reality--which no doubt in itself entails a whole bunch of metaphysical assumptions: but so be it!

Rather, if one understands what science really is, one sees that we are not choosing what to believe, we are choosing our objectives for forming a belief in the first place. The rest follows with little cognitive disquiet.

Sure, if we choose science--but we certainly are not required to choose science! Because choosing science often leads to ethical disquiet if not cognitive disquiet.

So a theory that is "falsified" within one accuracy target might be "confirmed" within some looser target (history is replete with that)

The Titius-Bode pattern is an example easily understood even by nonphysicists within the history and philosophy of science. Compared to spacing patterns chosen from a uniform random distribution, the TBL pattern wins hands down. But that still doesn't satisfy a lot of other people--so they insist that random simulations should draw from a logarithmic probability distribution or that the TBL should be compared against other (nonrandom) ideas, like the "law" of increasing differences.

So is the TBL confirmed or falsified? That depends on one's choice of a null hypothesis--there is no clear cut answer.

But lest we get too hung up on the issue of accuracy, note that science is about more than just the error in the prediction. It is also about organizing information, and suggesting new avenues for study. A good theory does this as well, and if it does it well, larger inaccuracies become tolerable. We also heard that sometimes it's not obvious what aspect of the theory is causing the problem, perhaps some idealization was made in a theory that is otherwise quite good.

Again, the TBL is a good, easily understood (compared to "CDM", whatever that is) illustration of this very point. The TBL does organize information--it was instrumental in predicting the locations of at least two planets--and it does provide constraints on theory, therefore suggesting new avenues of study.

Yet the TBL is very "messy". It's not that accurate (the worst residual errors range from up to 10% to 20%, depending on model chosen).

As long as a theory is generating the right questions to lead to advancement and progress, it is doing its job.

[Ken G]

And so why shouldn't people "turn to other means" to obtain the truth?

My issue is not that they do that, it's that they denigrate the conclusions of science in the process. Some truths can be "chosen", but it changes the nature of the truth, and many people seem to want it both ways.

The rebelliousness you mention is mainly because of the perceived ethical consequences that ensue--not from science per se--but from the perceived ethical consequences of scientifically inspired metaphysical systems such as atheism, naturalism, and materialism.

Right, but each of those is a philosophy with no direct connection to science, when one recognizes the difference between science and philosophy. That's part of the point of what we're trying to establish in this thread.

And since theory is underdetermined by the evidence, given the choice, should not one choose the more ethical alternative???

Ethics are never any part of the "choice" involved in scientific theory. What choice do you think you mean here?

Naturalism says that what is real is just what the scientists say is real;

You are again confusing science for philosophy. Scientists do not "say things are real", they offer theories that succeed at what they are intended to succeed at. [edit:] If one wants to participate in that success, one should know where it comes from, but there is no necessity to "buy off" on a particular belief system.

Nevertheless, naturalism looks only to science to say what is and how we know what is.

It sounds like you are again confusing science and philosophy. There is no need for an "ism" in doing science.

So what makes science different from other epistemologies is not merely the connection between theory and experience--no, it is the connection between theory and a certain kind of experience having to do with nerve endings at the interface between one's body and the rest of reality--which no doubt in itself entails a whole bunch of metaphysical assumptions: but so be it!

Sorry, that doesn't sound like any science I know.

Sure, if we choose science--but we certainly are not required to choose science! Because choosing science often leads to ethical disquiet if not cognitive disquiet.

Sorry, once again not the science I've ever heard of. You are mistaking unethical things done in the name of a misguided idea of what science is. Ethics are completely different from science-- science is moot on the issue. It is not what science is for.

So is the TBL confirmed or falsified?

Neither. It is not even a law, it is just numerology. Efforts to explain it would be the road to a law. Science does not simply notice patterns, it attempts to find a reason for them.

The TBL does organize information--it was instrumental in predicting the locations of at least two planets--and it does provide constraints on theory, therefore suggesting new avenues of study.

I am not aware of it ever being used to predict the location of any planet.

[DrRocket]

...
You are again confusing science for philosophy. Scientists do not "say things are real", they offer theories that succeed at what they are intended to succeed at.l If one wants to participate in that success, one should know where it comes from.

It sounds like you are again confusing science and philosophy. There is no need for an "ism" in doing science.

Sorry, that doesn't sound like any science I know.
Sorry, once again not the science I've ever heard of. You are mistaking unethical things done in the name of a misguided idea of what science is. Ethics are completely different from science-- science is moot on the issue. It is not what science is for.
Neither. It is not even a law, it is just numerology. Efforts to explain it would be the road to a law. Science does not simply notice patterns, it attempts to find a reason for them.
I am not aware of it ever being used to predict the location of any planet.

I completely agree with your stance. But in one respect I am a bit puzzled. How can anyone confuse science with philosophy ?

In Dreams of a Final Theory Weinberg has written an entire chapter entitled "Against Philosopohy", and a nice piece of logical thinking it is. And I find myself in agreemeht with Cicero's statement that "There is no position so ridiculous that it has not been held by some philosopher."

This is not to say that there is not value in the thoughts of a select few of he better philosopohers, but the value in advancing scientific research is nil. And, no doubt someone will point out a great advance or two made by someone like Descarte, but I will point back and note that the advance was made in a time when one person could understand enough to advance more than one discipline and Descarte's contributions to science were made while he acted as a scientist. Coordinate geometry has been useful to scientists. "I think therefore I am" has really not produced much that is tangible.