Talk:Falsifiability/Archive 6

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Just to say that Mayo's book is entirely written in the context of Lakatos' view, which ignores Popper's definition of falsifiability. Mayo's view of Popper's contribution is entirely based on Lakatos. This fact makes it unlikely that the book will make any connection with falsifiability. In fact, a simple search in the book shows that the term "falsifiability" is not even used once. I have little doubt that what the book says about probability in science is very interesting and valid. I also have little doubt that there is a connection to be discovered with falsifiability and it might not be hard to discover it. However, what is needed is a high quality source that makes this connection explicitly. Otherwise, we could also include a section on artificial intelligence, which is also seen by many as evidence that induction can be formalized. For example, I am a fan of chess and one chess expert said that (I paraphrase) "AlphaZero only used observations, no laws, to learn to play chess". Of course, it does not justify that we add a section on artificial intelligence in the article. What is needed is a reliable source that analyze explicitly this claim in the context of falsifiability. Mayo's book also appears to claim that progress in science can be explained as the result of statistical observations.

Both of these subjects would better fit in the article on the problem of induction. In the case of Mayo's book, I refer to the claim that is mentioned above. I don't see Mayo's book used in this article, nor does it say anything about artificial intelligence. Personally, for what it is worth, I don't believe in any form of the induction principle. I don't deny the fact that AI and also typical applications of statistics seems very much evidence that there is some formal induction principle and also that AI and statistics are valid approaches in practice, but this is the same old story that has always been at the basis of the problem of induction: it does not prove that there is a formal induction principle. Anyway, again, if I am wrong, then this should be included in the article on the problem of induction. More than that, the authors should receive the Fields Medal, not just a Lakatos Award, because changing the foundation of logic is not a small task. I am being sarcastic, I admit. Clearly, the solutions of Hume, Kant and Popper, all of them, are way more reasonable than believing in an inductive logic. Dominic Mayers (talk) 17:01, 19 April 2020 (UTC)

The Mayo book was not the only clue that I provided; I also suggested Gelman and Shalizi's paper, which cites Mayo and discusses falsifiability of statistical models:

Our view of model checking, then, is firmly in the long hypothetico-deductive tradition, running from Popper (1934/1959) back through Bernard (1865/1927) and beyond (Laudan, 1981). A more direct influence on our thinking about these matters is the work of Jaynes (2003), who illustrated how we may learn the most when we find that our model does not fit the data—that is, when it is falsified—because then we have found a problem with our model's assumptions. And the better our probability model encodes our scientific or substantive assumptions, the more we learn from specific falsification. (pp. 18–19)

I also suggested checking Google Scholar to see who else has cited Mayo and what some of them say about falsifiability (e.g. in the context of statistical modeling). I just did that myself and found several relevant references, including, for example: Dienes, Zoltan (2008). Understanding Psychology as a Science: An Introduction to Scientific and Statistical Inference. New York: Palgrave Macmillan. ISBN 0230542301. OCLC 182663275. {{cite book}}: Invalid |ref=harv (help) Dienes wrote:

The most natural application of Popper's thought to probabilistic hypotheses may in fact be a version of likelihood inference (see Chapter 5 for explanation of likelihood inference; and Taper and Lele, 2004, Chapter 16, for discussion with respect to Popper). For Popper the relative likelihood would not entail that hypotheses are supported in the sense of having increased probabilities (they are not inductively supported) but only in the sense that they are corroborated. (Popper himself was comfortable using the term 'support' provided by relative likelihood; see e.g. Section 2 of the Addenda to his 1963/2002.) [...]

Popper often said that the falsifiability of a theory can be discerned from its linguistic or logical form. For example, the statement 'all x are y' is on the face of it falsifiable because finding a single accepted case of an x that was not a y would show the statement wrong. On the other hand, the statement 'some x are y' is not falsifiable, given we do not have access to all the x's. No matter how many xs we find that are not y, 'some x are y' could still be true. But then to what extent are theories in psychology falsifiable? We saw in the last section that many claims in psychology are probabilistic, that is, consist of statistical hypotheses ('The population means for the two groups differ by more than 5 units'). Statistical hypotheses are not of a form like 'all x are y', but we may, as Popper suggests, be able to set up conventions to apply a falsificationist methodology (see Chapters 3 and 5). What is really important is falsificationist attitude, rather than syntactic form.

A statistical hypothesis on its own is an impoverished psychological theory. To be satisfactory, a statistical hypothesis should be strongly motivated by a substantial theory, that is, a unifying idea from which many predictions could be drawn. For example, one might use the idea of cognitive dissonance (a substantial theory) to predict attitude change in a particular context (a specific statistical hypothesis). Often in psychology (and more generally in cognitive science, Boden, 2006, and in the life sciences, e.g. Bechtel, in press) substantial theories take the form of mechanistic explanation, that is by postulating a mechanism by which something is achieved. The specification of a mechanism may consist not of lists of propositions but of analogies or models (cf. Giere, 1999). While such representations can be far from the linguistic structures (e.g. 'all x are y') the logical positivists started with, and Popper continued his thinking with reference to, one can still apply a falsificationist attitude to such theories. There are consequences of mechanisms working in a certain way, consequences that may show in behaviour, reaction times, brain imaging or lesion studies. Despite what Popper himself has often said, the application of his philosophy, or the spirit of it, does not depend on psychological theories having a certain syntactic structure (cf. Lakatos, 1978). The better one can specify a mechanism such that possible observations can refute it, the more quickly we may learn about the actual mechanism in nature. (pp. 25–26)

What Dienes discusses here in brief—the expansion of scope of falsifiability beyond (what he here calls) syntactic structure to, e.g., nonsyntactical/nonsentential models (studied by statisticians such as Gelman and Shalizi and philosophers of science such as William Bechtel and Ronald Giere, whom Dienes cited)—is what I was pointing toward. Biogeographist (talk) 18:33, 19 April 2020 (UTC)

It seems that, just like Popper complained about so often and just like Lakatos never appreciated, you do not seem to appreciate the distinction between falsifications and falsifiability, that is, you don't seem to appreciate the fact that, in Popper's view, there is a fundamental distinction between the logical part of science and its methodological part. In the case of Lakatos, isn't it interesting that he wrote an entire book about Popper's philosophy and claimed to help explain it to others, but without even giving any definition for falsifiablity? In this book, he immediately goes to the concept of falsification. The 5 decisions mentioned by Lakatos in this book to explain what he call Popper's falsificationism are all in the context of falsifications. When I saw this situation, my first thought was that maybe Popper was wrong to insist on this distinction and that all important concepts can be covered without it and that is what Lakatos and others understood. But, if that was the case, Lakatos should have been explicit about this, but he does not say anything about this, nothing, just as if he did not have the brain connections required to even only begin to conceptualize this distinction. In the same way, not surprisingly, given that the book follows Lakatos, Mayo does not consider at all this distinction. BTW, I said that the term "falsifiability" does not appear in the book, but certainly the term "falsification" is used in the book. The book even considers three of the five decisions that Lakatos says must be taken in a falsification. Also, I retract from my position that the book cannot be used in the article. It certainly can, just as Lakatos' book could be used. However, it will only fit in the sections before the definition of falsifiability, perhaps in the section Falsifiability#Away_from_dogmatic_falsificationism.

I am not going to argue here why the distinction is important. This is perhaps the most important aspect of falsifiability and yet one that is often misundertsood, so much that Popper said that an entire literature was written because this distinction was not understood. I would only say a few words about the importance of this distinction. I don't see that we can pretend to account for what is going on in science without making that distinction. We can say a lot about science with the impression that we covered all aspects, but without this distinction, something important is missing. This distinction is ubiquitous in science so much that we might not recognize it, just like the eyes cannot see themselves. The reason why this distinction is ubiquitous and yet not recognized is also the reason why there is rarely a debate about the empirical nature of basic statements.

Dominic Mayers (talk) 19:47, 19 April 2020 (UTC)

In relation to the issue of this article's scope, I don't see how it matters that in Popper's view, there is a fundamental distinction between the logical part of science and its methodological part. (Perhaps the fact that some philosophers of science have found this distinction to be unimportant is related to the fact that not all philosophers of science have been enthusiastic about Popper's three worlds ontology: for example, philosopher of science Mario Bunge was a detractor, as he was of Popper's philosophy of science in general, although Bunge agreed with the general criticism of inductivism.) The subject of this article is not Popper's philosophy of science; the subject of this article is falsifiability and falsification in all philosophy of science and epistemology. So the fact that not all philosophers of science have made exactly the same distinctions that Popper made does not mean that other aspects of falsifiability and falsification that Popper did not consider should not be discussed in this article. Biogeographist (talk) 22:23, 19 April 2020 (UTC)

Biogeographist wrote (Perhaps the fact that some philosophers of science have found this distinction to be unimportant is related to the fact that not all philosophers of science have been enthusiastic about Popper's three worlds ontology: for example, philosopher of science Mario Bunge was a detractor, as he was of Popper's philosophy of science in general, although Bunge agreed with the general criticism of inductivism.) I have read many of Popper's book, but I never been inclined to read about Popper's three worlds. I just did now, because you somehow suggested that it was a basis to understand the distinction between the logical part of science and its methodological part or why it is important for Popper or, at the least, that it is somehow related to it. I found the text interesting, but I would have a hard time to make the link with the question that interest us now. If I try, I would say that the logical part of science should go in the third world, but I don't know if the methodological part should go in the second or third world, though I am inclined to put it as well in the third world, as a creation of the human mind. In any case, I don't see why people who don't like Popper's three worlds should necessarily find unimportant the distinction between the logical part and the methodological part of science. I think it is quite straightforward to distinguish between activities that are based on mathematical logic from other mental activities and that's all what there is in Popper's distinction between the two parts of science. (Well, there is a bit more, because the empirical basis can always be moved "deeper" if needed, but it's not a big issue.) The fact that you try to put in the same basket this simple distinction made very early by Popper with a distinction that he made much later about three worlds makes very clear that you do not see the pertinence of this distinction in the philosophy of science and certainly do not understand what I meant when I wrote that it is ubiquitous in science. You should try to understand it.

As far as the remainder of the comment is concerned, I agree with what I understood. I would have formulated it differently than you, but I think you meant that it's not because Popper introduced falsifiability that only Popper's opinion is in the scope of the article. I obviously have to agree with that. I also agree that falsification is obviously sufficiently related to falsifiability to be in the scope. In fact, a significant part of the paper is only about falsification, as a way to smoothly introduce falsifiability, which only involves the logical part of science (except for a necessary link with factual observations, but the important point is that this link does not require that we consider any falsifications). I am disappointed that we still do not have a link between falsifiability and probability theory. A link between falsification and probability theory is in the scope, but if it takes too much space it will be disproportionate. There should be a separate article for the subject then. There is a natural balance to respect. We do not want to exclude any point of view that is well sourced, but at the same time Wikipedia should have a modular structure where each article has a natural focus. So, excluding a recent view point in philosophy of science that is based on falsification would be POV pushing, but if this other view point takes a lot of space, we should say the essential in the current article about it and the rest of it should be in another article by itself. Even though it is not restricted to Popper's view, this article is definitively about falsifiability as introduced by Popper, the logical criterion, not the process of falsification in science in general.

Otherwise, it is as if you felt that falsifiability does not deserve an article by itself and, worst, the technical name falsifiability, which was definitively introduced by Popper, would be used with a different meaning. The reason for that would be only that you think Popper's view, which is that we must distinguish between the logical part (where falsifiability lies) and the methodological part of science, is not important and, consequently, falsifiability as he clearly defined it is also not important.

Dominic Mayers (talk) 00:51, 20 April 2020 (UTC)

I did not say that Popper's logical/methodological distinction is the same as his three worlds ontology; I meant to imply that in both cases when other philosophers ignore or dispute such ideas of Popper, they have their reasons for doing so. I have a basic understanding of Popper's logical/methodological distinction, but I don't see how it is "ubiquitous in science". Certainly I agree that reasoning is ubiquitous in science, but that is different from asserting that Popper's distinction exactly as he understood it is ubiquitous in science—and if Popper's distinction is so ubiquitous, then why didn't Lakatos, for example, appreciate it, as you said above? Anyway, that issue is not directly relevant to improving the article, and talk pages are not a forum for general discussion of the subject, so we should drop the off-topic banter. I don't intend to do any substantial work on the article myself, so I don't have any other suggestions at this time.

Again, remember that there are a number of different terms that redirect to this Wikipedia article, and the disambiguation page on Falsification points to this article for the science-related meaning of falsification. So currently, according to all the redirects pointing to this article, the scope of the article should be wider than Popper's view, and it is, to some extent. All I have done is suggest what could added. Biogeographist (talk) 02:31, 20 April 2020 (UTC)

I don't think that we should make any fundamental change to the scope of an article that would compromise the natural modular structure of Wikipedia because of redirects and disambiguation pages. It's the opposite: the redirects and disambiguation pages can easily and should be adapted to respect the natural modular structure of Wikipedia. So, we can put aside these redirect considerations when we consider the natural scope of the article.

On the other hand, the discussion that we have regarding what is falsifiability and the fact that it lies at the logical level is very fundamental to determine the scope of the article. You wrote Popper's distinction exactly as he understood it as if it was different from the usual distinction that I believe you understand. Popper is simply referring to the distinction that every one who knows about logic as a language to express mathematical formula, together with inference rules, etc. can understand. The fact that it is ubiquitous is expressed in this this footnote, which is sufficiently important that Thornton expanded on it in his article about Popper.

What is ubiquitous is the fact that an empirical basis is implicit in the logical structure that is used. Scientists are not thinking "Ok let's define an empirical basis", but they have one implicitly and it's ubiquitous. To appreciate the importance of the distinction and that falsifiability must be defined at the logical level, you must ask yourself if this very natural process that is ubiquitous in science could work if theories did not respect falsifiability as defined by Popper. Clearly, the definition of falsifiability can be at the logical level and must be at that level to avoid all the issues with falsifications. That's why the distinction is important. Dominic Mayers (talk) 03:17, 20 April 2020 (UTC)

Yes, I agree that what is ubiquitous is the empirical basis of the logical structure, but Popper defined that logical structure (and falsifiability) in sentential terms, and the logic need not have that structure. That's what I mean by Popper's distinction exactly as he understood it. As Dienes put it in the quotation above: "While such representations can be far from the linguistic structures (e.g. 'all x are y') the logical positivists started with, and Popper continued his thinking with reference to, one can still apply a falsificationist attitude to such theories. [...] Despite what Popper himself has often said, the application of his philosophy, or the spirit of it, does not depend on [...] theories [or models] having a certain syntactic structure". That's all that expanding this aspect of the scope of falsifiability would mean. You may disagree, but that's all I'm trying to say. Biogeographist (talk) 10:39, 20 April 2020 (UTC)

I had already agreed that we want a generalization of falsifiability that considers probabilistic statements. I was even very specific. By such a generalization I explained that I meant a generalization of the definition itself, not only an argument + extra decisions to allow the current non probabilistic definition to apply to probabilistic statements. Of course, this implies that basic statements can be statistical statements and thus a generalization of what we mean by the empirical basis.

However, there are two basic ingredients in Popper's concept of falsifiability that I suggest define the general concept of falsifiability and thus the natural scope of the article. The first ingredient is a distinction between the logical side and what Popper calls the methodological side, because that is the key ingredient that allows a criterion to be applied on the logical side without having to determine whether an observation is factual or not. That's very crucial, because we have now the platform to define a criterion without the complication of falsifications (or of its generalization in the statistical case). The second ingredient is of course a criterion that is similar to falsifiability that is given in terms of the empirical basis.

The excerpts that you presented criticized the fact that Popper's notion of falsifiability depended on the syntactical structure (+ inference rules), which is true. Popper might even have said that the empirical basis itself can be determined syntactically, but of course, the context is that the logical framework with its interpretation is already fixed. Popper never said that this logical framework and its interpretation can be determined using only syntactical considerations. This would have been so ridiculous to say. It makes no sense to interpret Popper in this way. More importantly, there was no indication at all of an intention to keep the two main ingredients that I mentioned above, no appreciation of the central points of Popper's falsifiability. So, I got the filling that these excerpts were only picky and did not really shown a minimal effort to understand Popper and were only trying to justify a study of falsifications following the approach of Lakatos (I only mean here - ignoring falsifiability and directly considering the problems with falsifications), only being more clever in the study of falsifications or of errors as Mayo call them.

Dominic Mayers (talk) 15:13, 20 April 2020 (UTC)

I don't think it's true that I haven't tried to understand Popper and I don't see how that claim is relevant. I have to conclude that I'm not communicating well. I'm sorry that I was not able to provide any helpful information. Good luck. Biogeographist (talk) 16:35, 20 April 2020 (UTC)

The entire comment was not referring to you, but to these excerpts, which I assumed are not from you. An imaginary critic might reply that it remains irrelevant whether the authors of these excerpts tried to understand Popper or not. But, for me, the expression "do not show effort" only refers to what is seen, not to the actual efforts. Indeed, the actual efforts are not my concern. Moreover, I see this in its context: Popper said that an entire literature did not understand this point. So, I am not judging the people. It remains a state of affairs that there is a misunderstanding. You suggested that it is because it is not important. I present the other viewpoint: it's important and the texts that I see do not show an effort to understand Popper. Dominic Mayers (talk) 17:15, 20 April 2020 (UTC)

Just to add that I read more carefully these excerpts and I see now that the basic idea is that we can still have the attitude of a critical rationalist even if we cannot show formally that the hypothesis that is being studied is falsifiable. Well, I think Popper said that critical rationalism can be applied in metaphysics as well. You can start with a problem that has nothing to do with a scientific question and apply a critical attitude when trying to answer it. So having a critical attitude does not mean that you are dealing with a scientific theory. I have a hard time imagining a scientific hypothesis that cannot be written in a way that shows its falsifiability. If it can make predictions, it can be falsified. I will have to read the book to see what is meant. Dominic Mayers (talk) 18:28, 20 April 2020 (UTC)

Dienes goes on in the next paragraph after the part that I quoted to say that perhaps many theories (or models) could be rewritten as you just described (p. 26). So Dienes makes the same point that you just made. Something like this could easily be mentioned in the article: that falsifiability can apply to theories or models even if they do not have the explicit sentential structure that Popper described. This is not currently clear in the definition of falsifiability in this article, unless I overlooked something. Biogeographist (talk) 18:59, 20 April 2020 (UTC)

Good point ! It's not because we don't know how prove falsifiability that it's not falsifiable. That's not specific to falsifiability. It's true for every statement that we don't know how to prove. In itself, that's not very interesting, but I am sure the examples that Dienes has must show that it is a practical point to keep in mind in the case of falsifiability. Dominic Mayers (talk) 19:20, 20 April 2020 (UTC)

But note that the article as it is now does not say much about syntactic requirements for falsifiability. It does not say that the statement has to be of the form For all X, ... or anything like that. It just says that it must be contradicted by some basic statements (and be consistent with some others). Nevertheless, the basic point remains valid. It could be that we have a falsifiable theory and we don't know how to show that it is falsifiable. Dominic Mayers (talk) 19:29, 20 April 2020 (UTC)

The current account in the section "The demaraction problem" of the logical positivism and the much related, if not the same, logical empiricism movement as well as their relation with the Vienna circle is not verifiable and, in any case, too much detailed for the need of the article. Here is an excerpt from [1]:

The term ‘logical empiricism’ has no very precise boundaries and still less that distinguishes it from ‘logical positivism’. It is therefore hard to map. ‘Logical empiricism’ here includes three groups: (1) the Vienna Circle, here taken broadly to include those who were part of various private discussion groups, especially that around Moritz Schlick, and also the members of the more public Ernst Mach Society (Verein Ernst Mach), (2) the smaller, but perhaps more influential Berlin Society for Empirical Philosophy (later called the Berlin Society for Scientific Philosophy), and (3) those influenced by or who interacted with members of the first two groups and shared an intellectual kinship with them. Besides Vienna and Berlin, there were important centers of the movement in England, France, Scandinavia, at several universities in the U.S., and even China. This characterization includes thinkers who disagreed with doctrines espoused by members of the original groups and even some who defined themselves in opposition to the movement. This results in a vague boundary, but it suffices to identify a movement in which a large number of able philosophers self-consciously participated and to distinguish logical empiricism from other movements.

It does not, however, distinguish logical empiricism from logical positivism, and it is doubtful that any principled such boundary can be drawn along doctrinal or sociological lines (Uebel 2013). Usually when distinctions are drawn, ‘logical empiricism’ is the wider term.

— https://plato.stanford.edu/entries/logical-empiricism/#MapMov

Later in the same source, we have the following:

Since Newton the most paradigmatic examples of empirical science were those claims, usually quantitative ones, that were properly inferred from or appropriately confirmed by experience. Speaking very informally, these are the ones that we have good reason to believe or at least better reason to believe than the available alternatives. The problem, of course, is to specify the form of proper inferences, the form of an appropriate confirmation relation, and/or the structure of good reasons. The task is daunting, but logic in a suitably broad sense seems to be the right tool.

— https://plato.stanford.edu/entries/logical-empiricism/#MapMov

So, the essential point of the paragraph is well verified, only the description of the movement itself was not verifiable. The essential point is found in the terms "inference", "structure of good reasons" and "logic in a suitable broad sense". Popper shared the view that the claims that we accept and work with are "the ones that we have good reason to believe or at least better reason to believe than the available alternatives." However, in contradistinction to the logical empiricism movement, Popper rejected entirely the role of any form of logical inference or use "of good reasons"^[1] to explain how claims gain their validity.

I recall the basic of the article for completeness. Popper did not have a very good replacement for these explanations of why we believe in these claims, which we call the laws of science. His idea was that the conjectures and our belief that they are valid come from an organic (thus out of human reason) process. So, we have (non logical) conjectures and (logical) refutations. It's not hard to understand why Lakatos and many others criticized Popper saying that he did not really propose a solution to the main question of the philosophy of science. The entire focus of Lakatos and of many others was on showing Popper wrong on that respect. However, Popper was also interested to another very important problem, the demarcation problem, and he had an important extra, perhaps more subtle, ingredient in his philosophy of science: falsifiability. This ingredient does not show up in the expression "conjectures and refutations". In particular, it is certainly not included in the "refutations" part of the expression. Falsifiability is far from being a complete explanation for our beliefs in laws and it's not its main purpose. Maybe for this reason, it was ignored by Lakatos. However, falsifiability is nevertheless, if we consider how broadly it was accepted as such, a good criterion to demarcate scientific theories from non scientific ones. For some reason, apparently, Lakatos focused entirely on the problems of falsifications, the fact that they could not in themselves explain why we pick one theory over another, etc. and presented these problems as if they applied to the totality of Popper's philosophy. He ignored falsifiability, which is the subject of this article.

In brief, Lakatos ignored that Popper proposed a solution to one of the most important problems of the philosophy of science, the demarcation problem, and instead focused on the fact that he did not like Popper's solution to a different problem: how to explain the growth of scientific knowledge.

But why Popper claimed that falsifiability was also the solution to the induction problem? Did he claim that? Maybe he only claimed that he had a solution to both problems, not that falsifiability alone was the solution to both problems. It is something to verify. In any case, he could have said that, because, he wrote, and he was right about that, that the main goal of the Vienna circle in trying to discover some induction principle was to defend science against metaphysics:

The logical empiricists were no mere bystanders. They, or at least the main leaders of the movement, were politically and culturally engaged. Even more important, this engagement was accompanied by the conviction that their cultures were incapable of the necessary reform and renewal because people were in effect enslaved by unscientific, metaphysical ways of thinking. Such ways of thinking might be exemplified in theology, in the racial hatreds of the day, in conceptions of property, and in traditional ideas about the “proper” roles of men and women in society. So to articulate a “scientific world conception” and to defend it against metaphysics was not just to express an academic position in the narrow sense. It was a political act as well; it was to strike a blow for the liberation of the mind. To articulate scientific methods and a scientific conception of philosophy was the essential first step in the reform of society and in the emancipation of humankind (Carnap 1958/2017, Creath 2009, Uebel 2012.

— https://plato.stanford.edu/entries/logical-empiricism/#MapMov

So, in a way, which might not have satisfied Lakatos, he made the induction problem irrelevant, because the main goal was achieved. He also accompanied falsfifiability with his organic view of science. So, perhaps he meant that the "falsifiability package", which comes with an organic view of science, fully addresses the induction problem.

Dominic Mayers (talk) 14:59, 21 April 2020 (UTC)

Yes, that's better. Biogeographist (talk) 17:35, 21 April 2020 (UTC)

I noticed that there are many publications in the reference list that are not cited in the article. The uncited publications would be better placed in a "Further reading" section. For example, there are "Further reading" sections in Confirmation bias and Philosophy of mind, two articles chosen at random from the featured articles list. Biogeographist (talk) 00:06, 19 April 2020 (UTC)

It's a good idea. These references were used in the past. In particular, there were references to Kant, Hume, etc. in the article. The historical content that used these references was removed (not by me), because it was felt that there was a lack of focus in the article. (However, I think it would make sense to have a brief historical note that points to main articles.) Dominic Mayers (talk) 00:34, 19 April 2020 (UTC)

I finished separating the references and further reading. Biogeographist (talk) 00:56, 22 April 2020 (UTC)

The section says

Much of the criticism against young-Earth creationism is based on evidence in nature that the Earth is much older than adherents believe. Confronting such evidence, some adherents make an argument (called the Omphalos hypothesis) that the world was created with the appearance of age; e.g., the sudden appearance of a mature chicken capable of laying eggs. This hypothesis is non-falsifiable since no evidence about the age of the earth (or any astronomical feature) can be shown not to be fabricated during creation.

This text is not verifiable and I don't see that there is much chance that it can be verified. It shows a misundersrtanding of a simple fact: Omphalos hypothesis <=> Standard theory + the added statement that God created this recently, including the deceiving evidence that it is much older. This implies that the Omphalos hypothesis is falsifiable, because the conjunction of a falsifiable theory with any other statement is falsifiable (or logically false, but this is not the case here). What we could perhaps say is that no amount of scientific evidence will ever be able to refute the added statement by itself. But, the difficulty is that this added statement, in its strongest interpretation, implies the standard theory as far as the possible (deceiving) observations are concerned. We can certainly find a part of the statement that is not falsifiable, but that is also true for Newton's theory. So, it gets complicated and we are not going to do original research here.

(Though, my personal view is that we can say that the added statement is "non falsifiable" in the following sense that, even if we consider that it implies the standard theory, it remains that it does not add any degree of falsifiability to it. However, it would be more accurate to say that it can be discarded using the Occam's razor principle: between two theories that have the same predictions (the same degree of falsifiability), we prefer the theory that is simpler. But I am not aware of a discussion on the subject in a reputable source. The difficulty is that there is not a unique notion of simplicity. So Occam's razor is not clearly defined. However, here we only need the notion that a statement A is simpler than a statement A & B.)

We need to find reputable sources on falsifiability that makes a point on young-Earth creationism. In the long text below, we can see that some might view sentences from the skeptic literature such as "no amount of scientific evidence will ever be able to refute it" by Martin Gardner as meaning that it is non falsifiable, but Martin Gardner has written nothing or almost nothing about falsifiability and therefore is not a valid reference on the subject. I think we have some chance to see a reputable reference on the subject in the context of simplicity, which is an extra criteria that goes beyond falsifiability per se. The article does not currently cover the aspect of simplicity, but there are reputable sources on the subject in relation with falsifiability, even from Popper. Dominic Mayers (talk) 04:17, 30 March 2020 (UTC)

arbitrary section break

(Sorry for the section break, but the limited tool I'm using during coronavirus isolation doesn't have the ability to add more lines to the section above).

Martin Gardner was not an academic (though he had very frequent contact with academics), and his interest in philosophy was in areas probably very different from your interests -- but he knew more about the Omphalos hypothesis, and how to validly apply basic scientific principles to evaluate it, than you appear to do. In any case, if you say that the Omphalos hypothesis is falsifiable based solely on your personal philosophical reasonings, while Martin Gardner says that "no amount of scientific evidence will ever be able to refute it", then your assertion might be considered to be "original research" contrary to available sources... AnonMoos (talk) 15:39, 2 April 2020 (UTC)

I am not oppose to Martin Gardner position. He wrote "[It is] so in accord with geological facts that no amount of scientific evidence will ever be able to refute it." The phrase "no amount of scientific evidence will ever be able to refute it" should not be taken out of context. It is a consequence of "so in accord with geological facts". The logic here is clearly that an evidence against it would have to be against geological facts also. But, in the concept of falsifiability, we don't care about facts. The sentence "this bird weigh two thousands tons" can not be a fact at all and yet it can be used to show the falsifiability of "All birds weigh less than one thousand tons". Therefore, we cannot use this sentence to claim that Martin Gardner advanced the position that the Omphalos hypothesis is non falsifiable. You are trying to take some text out of context. As I said, Martin Gardner wrote nothing or almost nothing about falsifiability. Even if we find an isolated sentence somewhere by Martin Gardner or another skeptic that mentions falsifiability, I don't think it would deserve space in this article. Many books have been written to give one consistent notion of falsifiability (with variations as discussed in the article). An isolated sentence that would present a completely different and naive notion would have no weight in comparison to that. Dominic Mayers (talk) 18:58, 2 April 2020 (UTC)

You seem to know quite a bit more than I do about philosophy (I'm not in a position to evaluate whether or not you do), but the purpose of falsifiability is to be of service to scientists. If it's made so philosophically abstruse and esoteric that the vast majority of scientists can't apply it, then it defeats its own purpose.

And Martin Gardner never wrote at any length about falsifiability as a doctrine (as far as I know), but he knew quite a bit about scientific methods and reasoning, from working with scientists, from writing science popularization books (some, such as "The New Ambidextrous Universe", at a rather high level), etc. He had also read Philip Gosse's 19th-century book Omphalos, which is the fullest original statement of the Omphalos hypothesis. So what Martin Gardner says about the Omphalos hypothesis comes from a scientifically-minded person applying scientific reasoning to something he knew about in great detail. What you say about the Omphalos hypothesis comes from you applying your personal philosophical reasonings to something which you more or less know about from Wikipedia articles.

Also, when Martin Gardner said that the Gosse book "presented a theory so LOGICALLY perfect, so in accord with geological facts, that no amount of scientific evidence will ever be able to refute it", he meant that the Omphalos hypothesis was structured so that it was LOGICALLY completely impervious to all factual confirmation or disconfirmation, since it could agree with any physically possible set of geological facts. He very clearly did NOT mean that the Omphalos hypothesis was so supported by geological evidence that no further evidence could possibly discredit it (a very strange idea, if you think about it, which would make Gardner a creationist, something that he definitely was not.) AnonMoos (talk) 18:59, 9 April 2020 (UTC)

I did not interpret Martin Gardner as in your second interpretation. Now, you might think that your first interpretation, which you say is what Martin Gardner really meant, is the very definition of falsifiability, but even that is not clear, because you use the word factual. It remains a question of how we interpret factual. The way I use to avoid difficulty of interpretation like that is by having a few sources for the same content: the primary source and one or more secondary sources. This is why I don't only mention Popper, but I also look at the position of Lakatos on the subject of falsifiability and even of philosophers that have read both Lakatos and Popper to analyse the debate, etc. It's a lot of work. I am not opposed to new sources. But, this is an article on falsifiability and Martin Gardner did not write anything to my knowledge about it and doing our own interpretation would be original research. Dominic Mayers (talk) 19:26, 9 April 2020 (UTC)

Also, you defend a lot the reputation of Martin Gardner, but his reputation is not the issue here. He simply did not write on the subject. Otherwise, if he had written sufficiently on the subject, I am sure that we would find other sources for the same content, I mean that it would not be an isolated content, and we would not need to have this discussion. Dominic Mayers (talk) 19:55, 9 April 2020 (UTC)

As I said previously, his interest in philosophy, such as it was, was in quite different areas, but he did know enough about how scientists work that his decription of a hypothesis as being such that "logically...no amount of scientific evidence will ever be able to refute it" has a high degree of credibility. If the Omphalos hypothesis is falsifiable, then nothing is unfalsifiable, so that falsifiability has no interest to scientists... AnonMoos (talk) 11:37, 23 April 2020 (UTC)

I can agree with every thing you wrote, except the last statement. I don't see the link between this last statement and the first part of your comment. Again, you assume that this first part refers to falsifiability, but that requires a personal interpretation of it and thus is original research. Moreover, I reasonably argued that the most natural interpretation is not that the OH is unfalsifiable, but that the OH is the standard theory + a weird statement that adds no predictive power to this theory. Dominic Mayers (talk) 15:30, 23 April 2020 (UTC)

@Banno and Dominic Mayers: I see that there has been a lot of editing activity here recently. Don't forget that this article is the target of a number of redirects. In this edit I tried to make sure that the major redirect terms are in boldface in the lead or elsewhere, per MOS:BOLDLEAD & MOS:BOLDSYN & WP:R#PLA. If the lead is to be rewritten (as I think it should be, since in its current state it does not seem to summarize the subject well), keep in mind that the major redirect terms should appear in boldface.

Notice that Black swan problem and Black swan fallacy currently redirect here but neither of those terms appears in the current version of the article, nor is there any clear definition of those terms. This should be corrected if possible. I don't currently know enough about the subject to fix that issue myself.

Also, most of the philosophical part of this article is focused on mid-20th-century issues, especially from Popper's perspective. It would be great to have more discussion of the role of falsifiability in 21st-century thinking among those who are not disciples of Popper. Biogeographist (talk) 02:55, 18 April 2020 (UTC)

Regarding the black swan problem, if we accept the view of Black swan theory, it seems that it would be better redirected to the problem of induction. However, perhaps even better would be to ignore this until we find this expression used in high quality sources in a proper context.

Regarding the black swan fallacy, this is not discussed in the article and it would also be better discussed in the problem of induction or in Faulty generalization, because it is nothing else than the consequence of applying the logically invalid induction principle. Also here, I suggest to ignore this until we find this expression used in high quality sources in a proper context.

Regarding your concern that there is a bias toward Popper's perspective, this makes no sense. For example, about half the article presents Lakatos perspective, which was not at all Popper's perspective. I certainly edited the article without any bias at all. I was very innocent and only wanted to faithfully present the high quality literature on the subject. But, I already added a banner because important content is missing. For example, there should be a connection with probability theory and, I agree, if it is what you meant, that we should not restrict the sources to articles that focus on falsifiability. Indeed, I suspect that the best high quality sources, will be about probability in science in general and only make the connection to falsifiability in one paragraph or so. But, we need to find these sources. I mean, the generalization of the definition of falsifiability to the probabilistic case is not that hard, but still we need sources that make the connection explicit. Popper might have considered this generalization, but I have not seen this. Instead, he tried to reduce the probabilistic case to the non probabilistic definition, but that was a lot of work and I am not sure that he succeeded. Anyway, I agree, we don't care whether Popper did or did not do it, as long as it was done in a high quality source.

Regarding your weaker version In contradistinction, the logical positivism of some philosophers of the so-called Vienna Circle, especially Moritz Schlick and Friedrich Waismann, and the logical empiricism of A.J. Ayer, I believe that the original version is easily verifiable. It is well known that verificationism was the main thesis and the common view within the Berlin and Vienna circles. I have no idea where you find the idea that it was only isolated people within these circles. It's this weaker version that is hard to verify.

Dominic Mayers (talk) 10:02, 18 April 2020 (UTC)

After some investigation, I redirected Black swan fallacy and Black swan problem as suggested by Dominic Mayers above. There is also a new hatnote in Problem of induction indicating the redirect, and entries in Black swan were updated accordingly. Biogeographist (talk) 15:51, 23 April 2020 (UTC)

I have not studied the whole article in detail, but the lead and early sections still strike me as focusing on Popper's perspective. For example, the second paragraph read "Falsifiability was introduced by the philosopher of science Karl Popper..." and I changed "introduced" to "emphasized", which was changed back to "introduced". I don't really care what this article says, but the idea that Popper "introduced" the general idea of falsifiability seems to be totally the perspective that Popper himself promoted! It's his perspective, not WP:NPOV. His disciples may share that perspective, but not everyone does. Contrast that perspective with a historical perspective such as that found in: Niiniluoto, Ilkka (1984) [1978]. "Notes on Popper as follower of Whewell and Peirce". Is Science Progressive?. Synthese library. Vol. 177. Dordrecht; Boston: D. Reidel. pp. 18–60. doi:10.1007/978-94-017-1978-0_3. ISBN 9027718350. OCLC 10996819. As we shall see later, however, these problems [emphasized by Popper] were largely the same as the standard problems of nineteenth-century methodologists. Anyone who sets out to systematically compare Popper's methodology with the nineteenth-century conceptions of science will find out that many ideas of Whewell's and Peirce's, for example, are surprisingly similar to the epistemology and philosophy of science which Sir Karl has taught for many years in the conviction that it is new and revolutionary. (A similar observation on the relation between Berkeley and Mach was made by Popper in his, 'A Note on Berkeley as Precursor of Mach and Einstein'; see Popper, 1963, p. 171.) Rather than as the first and foremost representative of a new epoch, Popper can thus be regarded as an upholder of an important nineteenth-century tradition in methodology—to the extent, indeed , that Laurens Laudan has not unjustifiably claimed Popper to be 'probably closer to the nineteenth-century methodological tradition than is any other living philosopher' (Laudan, 1973a, p. 306). {{cite book}}: Invalid |ref=harv (help) Biogeographist (talk) 11:57, 18 April 2020 (UTC)

Regarding the Vienna Circle, it was not monolithic. The so-called Vienna Circle philosophers did not agree on everything, not even about the details of verificationism. The change I made may not be exactly correct, but to my knowledge what it replaced was not correct. These passages need good scholarly sources, and not just from a Popperian perspective. Biogeographist (talk) 12:03, 18 April 2020 (UTC)

This article is NOT about the general idea of falsifiability and the immediate context of the sentence Falsifiability was introduced ... makes very clear that we refer to the technical meaning of falsifiability. It would be exaggerated to require that we write The technical meaning of falsifiability was introduced .... What you suggest is perhaps that the difference between the meaning in common language and the technical meaning introduced by Popper is not important enough, but this is subjective. Or perhaps you suggest that the technical concept existed before implicitly and attributing it to Popper because he made it explicit is inadequate. This would also be subjective. We normally credit the person who makes the concept explicit. Dominic Mayers (talk) 12:26, 18 April 2020 (UTC)

I read quickly the reference that you provided and what I see is that the author argues that Popper did not discover new problems. For example, he wrote

Popper' s LSD 3 shows that already in the heydays of logical em-piricism he represented ideas and emphasized problems which later in the sixties have become subjects of intensive philosophical discussion. As we shall see later, however, these problems were largely the same as the standard problems of nineteenth-century methodologists.

— Ilkka Niiniluoto, Notes on Popper as Follower of Whewell and Peirce, abstract

But of course, we are not crediting Popper for the discovery of these problems. Even the notion of falsifiability is built upon already existing concepts. Nothing is really totally new. However, most author attributes falsifiability (in its technical meaning, of course) to Popper. Dominic Mayers (talk) 12:38, 18 April 2020 (UTC)

This being said, if you find a source that clearly gives the definition of falsifiability, not just the "general idea", before Popper, then it's great. Wikipedia should obviously credit this source. That's no problem at all. We don't care. What we want to avoid is to make things unclear by suggesting that falsifiability is nothing more than a general idea. The concept is very precise and well explained in many high quality sources. This is what this article is about. Dominic Mayers (talk) 12:50, 18 April 2020 (UTC)

Dominic Mayers wrote: What you suggest is perhaps that the difference between the meaning in common language and the technical meaning introduced by Popper is not important enough, but this is subjective. Or perhaps you suggest that the technical concept existed before implicitly and attributing it to Popper because he made it explicit is inadequate. Neither of those options is what I meant. I would not set up an opposition between "the meaning in common language" versus "the technical meaning introduced by Popper" or between a concept that was implicit before Popper and explicit after Popper.

Instead, the question I would point to is, given all the redirects to this article: What is the proper scope of this article on falsifiability/disprovability/refutability (and, to some degree, on verifiability, since that term also redirects here)? I can't answer that question by myself, of course, since the answer should be a product of consensus among editors, but I imagine the proper scope of this article would be: all the major conceptions of falsifiability/disprovability/refutability in epistemology and philosophy of science. There is no doubt that Popper would play a large role in such an article. But Popper's role should be placed within the larger context of the development of ideas about falsifiability/disprovability/refutability in epistemology and philosophy of science. This does not imply an "imprecise" discussion; it can be as technical as required, but it should be sufficiently broad in scope. Starting from the lead, it should be clear that scope of the article is this whole field of ideas, not just Popper's. There should be a way to write the lead that shows that Popper had a large role, but that the scope of the article is not all about Popper's role. In addition, it may be wise to put a History section after the lead, as is common in many Wikipedia articles. For example, Niiniluoto, cited above, mentioned William Whewell and Charles Sanders Peirce as precursors in some respects. Lakatos in "Falsification and the methodology of scientific research programmes" mentioned Whewell, Henri Poincaré, Gaston Milhaud, and Édouard Le Roy. Azarya Polikarov even mentioned Nicholas of Cusa and Blaise Pascal as precursors (and he said, correctly as far as I can see, that "Popper's opposition falsificationism (deductivism) – verificationism (inductivism) rests on an oversimplification") in: Polikarov, Azarya (December 1998). "A draft for unifying controversies in philosophy of science". Journal for General Philosophy of Science. 29 (2): 225–244. doi:10.1023/A:1008235508964. {{cite journal}}: Invalid |ref=harv (help) I imagine that other sources on this history could be found without too much effort.

I am not planning to work on this article myself, so I hope this feedback is helpful. It does not appear that you are currently getting much or any peer review of this article, so I expect you need all the feedback you can get. Also, I noticed that the pronoun "we" appears to be overused in the current version of this article: MOS:WE suggests rewriting to avoid such use of "we". Such usage is common in mathematics, so if you are accustomed to writing mathematical texts, you may need to keep in mind that such usage of "we" is discouraged in non-mathematical articles in Wikipedia. Biogeographist (talk) 15:27, 18 April 2020 (UTC)

Until we have sources (primary and secondary) of good quality that provide a technical definition of falsifiability in the context of philosophy of science, which is not attributed to Popper, we will accept the sources that we have, which attribute falsifiability to Popper. Putting aside slight variations on the original definition introduced by Popper, I am not aware of another definition. So, we are actually covering all the major conceptions of falsifiability. It just happens that there is only one.

Whether we should have an history section is a different issue. I would not oppose an history section. But, an history section, is not really a change in the scope of the article.

Finally, there is no universally accepted rule against "we" in articles. The only rules that I have seen only say to not refer to a specific group using "we", but even this is not a universal rule. In any case, I easily removed most of the "we". It was easy, because in all cases the "we" in the sentence did not have to refer to a specific group, which could have broken a rule. No rule were broken, but I made the modifications anyway. A typical example is "we cannot tell if ..." -> "it cannot be told if ...".

Dominic Mayers (talk) 18:46, 18 April 2020 (UTC)

When I read the article, it sounds like the article is referring to me, the reader, in many places where it says "we", which is often. And you admitted that such an effect is intentional, in the part of your response above that is commented-out: it's an imaginary group that can even include the reader. That is what MOS:WE (in Wikipedia's Manual of Style) says should be avoided. You can ignore the Manual of Style if you wish, since you are currently essentially the only person editing this article. But it's not a moot point; if it were, it would not be addressed in the Manual of Style.

I'm not going to spend time right now combing the literature to make a comprehensive list of non-Popperian approaches to falsification, but an example that comes to mind off the top of my head (like all the other sources that I cited above) is Deborah Mayo's approach, for example in her 1996 book Error and the Growth of Experimental Knowledge, which won the Lakatos Award and has been influential. Andrew Gelman and Cosma Shalizi, for example, have cited Mayo in their discussions of falsification of statistical models, e.g.: Gelman, Andrew; Shalizi, Cosma Rohilla (February 2013). "Philosophy and the practice of Bayesian statistics". British Journal of Mathematical and Statistical Psychology. 66 (1): 8–80. doi:10.1111/j.2044-8317.2011.02037.x. PMID 22364575. {{cite journal}}: Invalid |ref=harv (help) This is not something that Popper discussed, as far as I know, and the lead of this article as currently written is not broad enough in scope to accommodate it. Biogeographist (talk) 20:14, 18 April 2020 (UTC)

I will not respond to your reference to a text that I have commented out. Dominic Mayers (talk) 20:26, 18 April 2020 (UTC)

Do you care to provide a reason why? Why do you comment out parts of your responses? I honestly don't understand—I've never seen anyone do that before on a talk page, and I've been editing Wikipedia for years. If you wish to retract parts of your responses, you can delete them or strike them like this. Biogeographist (talk) 20:32, 18 April 2020 (UTC)

It does not matter why. If it is commented out, it means that it was not to be discussed. Dominic Mayers (talk) 20:38, 18 April 2020 (UTC)

OK. Keep in mind that commenting out your writing is not typical talk-page behavior, so you may have to explain this to other editors in the future as well. Biogeographist (talk) 21:19, 18 April 2020 (UTC)

I will read your reference Error and the Growth of Experimental Knowledge. My bet is that this source does not give a new definition of falsifiability. If it does, this new definition was ignored by most authors. Of course, the author might have a different view. So did Lakatos and many others. The article presents many views. In fact, it first presents Lakatos' view. Dominic Mayers (talk) 20:38, 18 April 2020 (UTC)

Regarding connection with probability and statistics, before you came in, there was already a banner to say that the article should be extended to make the connection with these subjects. It's going to be a natural extension, but I don't see that it will change the fact that Popper introduced falsifiability, etc. My personal view, which I already explained and is only useful as long as it might correspond to content sourced in reputable high quality sources, is that we should have a generalization of the non probabilistic definition. I never seen Popper suggests that and I don't know why. It should not be that hard to do this generalization. It's not very fundamental. (Instead, Popper tried to reduce the probabilistic case to its non probabilistic definition and this is much more complicated.) Dominic Mayers (talk) 20:52, 18 April 2020 (UTC)

I am just thinking now that maybe your reference Error and the Growth of Experimental Knowledge provides the generalization to the probabilistic case. I mean, I could be wrong when I suggested that it does not provide a new definition. In any case, I never expressed a problem in mentioning another definition as long as it is well sourced and have secondary sources, etc. In the case of the reference that you provided, it will not change the fact that Popper introduced falsifiability. It will just be a nice addition to the article, which was already invited using a banner. Dominic Mayers (talk) 21:06, 18 April 2020 (UTC)

I think we're understanding each other more now. I'm not suggesting a radical redefinition of falsifiability, just suggesting that the scope could be expanded, and you seem to agree. You may not be the right person to write a History section on Popper precursors, but at least the idea has been raised here on the talk page in case anyone wants to do it. Enjoy reading Mayo. You should read the Gelman and Shalizi paper too, if you're interested; it's great. If you get really curious, you could check Google Scholar to see who else has cited Mayo and what some of them say. Biogeographist (talk) 21:19, 18 April 2020 (UTC)

Well, maybe you did not see the banner that says clearly that the article should be extended to make the link with probability theory and did not understand what I meant by I suspect that the best high quality sources, will be about probability in science in general and only make the connection to falsifiability in one paragraph or so. But, we need to find these sources. This clearly says that, on my side, nothing has changed, except that now I have a reference that looks promising. Dominic Mayers (talk) 22:31, 18 April 2020 (UTC)

The banner is at the top of the Applications section. I suggest that the implications of Mayo and other writers who cite her about statistical modeling would not only lead to a special subsection in the Applications section (although that could be an option) but would also require some expansion of the scope of the discussion of falsifiability in the body of the article above the Applications section, e.g., expanding beyond basic statements and theories to models. That's a question to think about as you are reading. Biogeographist (talk) 23:49, 18 April 2020 (UTC)

Yes, I did not put the banner in the application section to mean that the missing content should be added in that section and the banner does not say so. I will move it at the top. It's only that I did not want to be too aggressive with that banner. Dominic Mayers (talk) 00:49, 19 April 2020 (UTC)

My concern is that the generalization of Popper's definition will be there, but not acknowledged as such. That would be problematic, a pity, because the link with a very precious definition that allows us to see many issues in the simple non probabilistic case would be missing. But, I am have not yet read the book, so we will see. Dominic Mayers (talk) 22:39, 18 April 2020 (UTC)

Regarding an history section, there is a possible issue. If you read the article, you will see that the connections with the Quine-Duhem problem and other historical issues is already covered and the only way to properly cover them, without being superficial and misleading, is to cover them in the correct context within the article. I am really concerned that people with only a superficial view on the subject do some cherry picking to push their personal view point and an history section could be the perfect section for them to do that. Dominic Mayers (talk) 22:54, 18 April 2020 (UTC)

Yes, the History section would need a title such as "History before the 20th century", to prevent it from accumulating a mini-POV fork of the same material that is discussed later. Biogeographist (talk) 23:49, 18 April 2020 (UTC)

I am considering Dienes as a source. Thus far it looks good, but I am just starting. From the book, I got an interesting example of a falsifiable statement: "there exist a unique color such that all swans have this color." It's an existential statement, but it can be falsified by the following singular statement: "Sam the swan is black and Jim the swan is white." The example in the book is not with swans and colors, but with models (instead of swans) and their parameters (instead of colors), but the idea is the same. Dominic Mayers (talk) 19:58, 25 April 2020 (UTC)

Well, the details are different, so that, if the space of parameters is infinite, a method to find successful parameters must be provided. Dominic Mayers (talk) 23:20, 25 April 2020 (UTC)

In the lead we have:

Falsifiability was introduced by the philosopher of science Karl Popper in his book Logik der Forschung (1934, revised and translated into English in 1959 as The Logic of Scientific Discovery), as an attempt to solve both the problem of induction and the demarcation problem.

This suggests that it was universally viewed (even by Popper) as an attempt. I don't want to play with words. So, I go directly to the point. I don't think we want to put judgments about the value of falsifiability in the lead. They would have to be attributed anyway and these attributed judgments fit better in the body of the article, in a proper context, after we have explained what it is. I propose to change it to

Falsifiability was introduced by the philosopher of science Karl Popper in his book Logik der Forschung (1934, revised and translated into English in 1959 as The Logic of Scientific Discovery). He proposed it as a solution to both the problem of induction and the demarcation problem.

The fact that Popper proposed it as a solution is neutral. Dominic Mayers (talk) 15:21, 29 April 2020 (UTC)

That's fine. Do you know that you don't have to explain small changes like this on the talk page? Generally people only use the talk page when they are asking about something that they are uncertain about, or when they are proposing large structural changes, or when there is some conflict of opinions, etc. When you know what you want to do, it is faster just to make the change and give the rationale in the edit summary, which also prevents the talk page from becoming cluttered with talk about small changes. Anyone who disagrees with the edit will start a discussion on the talk page. Biogeographist (talk) 17:41, 29 April 2020 (UTC)

I was not sure if it was a small change. Some people want a judgment to appear in the lead and it can become a big deal. Dominic Mayers (talk) 17:59, 29 April 2020 (UTC)

Just to point out that the expression falsifiability was almost not used in digitized (i.e., scanned) literature published before 1934. It is interesting that the largest increase in occurences of falsifiability coincide with the publication of Logic of Scientific Discovery in 1959, whereas the largest increase in occurences of the term provability coincide with the emergence of logical positivism. They might just be coincidences. Dominic Mayers (talk ) 20:57, 30 April 2020 (UTC)

Good idea to do a Google Ngram of the term. Although it is not directly related to this article, I checked the term "fallibilism" as well, and interestingly use of the term "fallibilism" starts around the same time as the term "falsifiability". But C.S. Peirce was a fallibilist, and he was dead a couple of decades before people started using the term "fallibilism"! Biogeographist (talk) 21:03, 30 April 2020 (UTC)

I added the ngram of falsificationism, just to be certain that, as expected, the notion of falsificationism emerged way after the notion of falsifiability. It started with the work of Lakatos that was influenced by Thomas Kuhn. Dominic Mayers (talk) 13:15, 2 May 2020 (UTC)

Excellent! Also note that the 1930s was the start of the institutionalization of philosophy of science in general; for example, the Philosophy of Science Association (PSA) was founded in 1933 and the journal Philosophy of Science was started by the PSA in 1934. The first edition of Peirce's collected works was also published for the first time in the early 1930s, which may explain why "fallibilism" did not begin to be used until then in the Google Ngram corpus. That spike in "provability" around 1920 (1910–1930) is very interesting. I searched for "provability" in Google Books and Google Scholar in texts from 1910–1930 and it appears that the term was especially used in bankruptcy law at the time, so one shouldn't infer that the spike of "provability" at that time period has much relation to philosophy of science. Biogeographist (talk) 16:24, 2 May 2020 (UTC)

@Dominic Mayers: Do you know the German equivalents of the terms? An Ngram of the German terms could be interesting. But I don't know how many German texts are in the Google Ngram corpus relative to English texts. Biogeographist (talk) 16:46, 2 May 2020 (UTC)

No, but it would indeed be interesting to look at the German terms. Yes, regarding provability, my original purpose when I added it was only to have a reference curve based on a similar word. I was not thinking that it would be connected with logical positivism. When I saw this pike, I thought maybe I missed this part of the literature on logical positivism, because I had never seen provability used in that context. I looked at verifiability, but it is a too common word. What is great about falsifiability and falsificationism is that, most likely, they are only used in the context of the philosophy of science. Dominic Mayers (talk) 17:01, 2 May 2020 (UTC)

When the title of the section "Problem of induction" was changed (see this edit) to "Falsificationism: a solution to the problem of induction", a need to clarify the role of this section in the organization of the article was, perhaps not on purpose, made apparent. What is the section about? If it is about the problem of induction, what is the purpose of the subsections on falsificationism? The simplest idea was that these subsections presented the solution. So this edit introduced some kind of global logic, which was missing.

However, it's not the way the article was truly organized. It's clear that the plan always was to present falsifiability as the solution that Popper proposed to both the demarcation problem and the problem of induction, not Lakatos' solution. Yet, this change of title suggested, perhaps not on purpose, that Lakatos falsificationism could also be presented as a solution. This was not an idea that I felt comfortable to reject quickly, because clearly falsificationism is in the scope of the article and we don't want to push one point of view over another. So, I worked hard these last days, reading Lakatos and other sources to see how I could add yet another subsection that would present more adequately Lakatos' view point. The main problem is that Lakatos is never presented as a separate solution. It is always presented as an improvement over and even a criticism of Popper's solution. It makes no sense to do that in the article before we present correctly Popper's solution. It's important to put "improvements" and criticisms in their proper context. But, Lakatos analysis of falsification (via dogmatic and naive falsificationism) by itself is indeed very useful to introduce Popper's solution. This is the basic idea behind the current organization of the article.

The new title "Falsificationism: a solution to the problem of induction" does not match with the current organization of the article. So, I will return it back to "Problem of induction." I have a rough idea how we can address the question "what is the purpose of the subsections on falsificationism?" The idea is that it is naive to say that "flipping the problem upside-down" was the solution. In fact, there is no flipping, because we had both falsification and (we thought also) confirmation (using induction). The problem is that we lost confirmations (not really, because we never had it) and we were left only with falsifications. In this perspective, the problems that we have with falsifications can be seen as more details about the problem that is created by the lost of an induction principle (the logical tool for confirmation or verification).

But, I retained the idea that Lakatos' approach should be better discussed in the article. It could be done in the section controversies, but it could also be done in a separate section, after the definition of falsifiability. We cannot put too much weight on this, because it's rarely presented as a solution, but more as an attempt and even more often as a criticism of Popper's solution. If there was a good consensus that Lakatos criticism was valid, we might not even have this article in Wikipedia. But the consensus is more that Lakatos misunderstood Popper. For example, Zahar wrote

Lakatos obviously ignores a fundamental truism: in order for a theory to be confirmed, whether dramatically or not, it ought by the same token to be falsifiable too. Confirmability and falsifiability are two complementary aspects of the same condition. Thus, Lakatos's methodology needs falsifiability just as much as does Popper's

The details are not important here. The general point is that in many occasions we can see that Lakatos never appreciated the role of Popper's logical falsifiability. His focus was always on practical falsifications and, indeed, he did, in my opinion, a better job than Popper in explaining the problems with falsifications. It's almost as if for Lakatos a criterion that exists essentially at the logical level (except for a link with actual observations that is based on an almost implicit agreement regarding the empirical basis) was a mistake. He always interpreted falsifiability as if the contradictions were only useful as potential falsifications, but they are useful independently of that, as Popper illustrated with his example where and apple is moving up from the ground. He always interpreted Popper's requirement that these contradictions must be known in advance as if one was required to say when his theory would be actually rejected from science, which is nonsense when we consider Popper's examples.

To be fair, when you read Lakatos carefully, you see that he has an interesting view about the philosophy of science. For Lakatos, a description of science must be corroborated in the history of science, looking at what is considered by most scientists as successes or failures. These historical successes or failures were, for Lakatos, the equivalent of potential falsifiers. A criterion that applies to a single theory at a time could not be used to reconstruct history, only a criterion on research programs could match history. This explains why he ignored falsifiability, except to criticize it as being inadequate because it was rejected by his historical criterion. The mistake is that Popper's falsifiability was never proposed to be used by itself as a way to reconstruct history. Popper made it clear that a theory does not have to be rejected from a research program even if it is falsified. Popper, in the same way as Lakatos, felt that it is important that a methodology corresponds to what great scientists have done in history and he was very much aware that many times falsified theories were not abandoned. So, putting aside Lakatos' misunderstandings about Popper's view, as Zahar points out, it's very hard to see what is the difference between Popper's and Lakatos' views. One key difference, which is pointed out by Zahar, is that Lakatos was seeking for a genuine logic of scientific discovered, but as far, as I can tell, he died before he had any success. His idea was based on Popper's definition of verisimilitude, but this definition was shown to be invalid the year of his death.

Dominic Mayers (talk) 17:59, 2 May 2020 (UTC)

Just to address the issue of the heading: Why not take "falsificationism" out of the "problem of induction" heading and add a new heading just before the paragraph that begins "Popper was aware that this basic idea would be subject to valid...", a heading that says something like: "Falsificationism: Lakatos on Popper"? Biogeographist (talk) 18:09, 2 May 2020 (UTC)

The same situation hold with the new title that you suggest as with the current title. The fact that it would be located just before that we mention Lakatos makes little difference. The title would still not match with the current content. It would bring us to the same situation. I think we should keep the overall idea of the section as it is: we first mention the basic idea of the solution, which is to rely on falsifications, then say that the problem of induction is still there, because there are problems with falsifications. It is simple. Yet, together with the section "The demarcation problem", it prepares perfectly the section about falsifiability, a solution that is free from these problems, because it does not use actual falsifications. In this plan, the section does not need to explain what is Lakatos' sophisticated falsificationism and how Lakatos criticized Popper, etc.

After we have presented Popper's solution, it makes sense to present Lakatos' view on Popper, his idea of sophisticated falsificationism with research program, etc., but I think it's natural to put this in the controversies section, because it's not at all a support for falsifiability. But, I do agree that we need to make some modifications so that it's clear that the current subsections that mention dogmatic and naive falsificationism fit perfectly well in the section. There is only a need to better explain the flow of the section and naturally the original title "The problem of induction" will be fine.

Just to add that a lot in Lakatos' work can be seen as complementary and in support of Popper's view. In fact, Zahar says that there is not much difference between Lakatos' and Popper's fundamental positions. My suggestion is that we continue as we do: we use every thing from Lakatos that is useful to explain falsifiability without changing headings or the natural flow of the article. The article uses already a lot, but perhaps we can use more. We might be able to use his notion of research program as a way to better explain Popper's solution, as a complement that supports it, instead of presenting it as a controversy. It is verified, because Zahar says that it fit well within Popper's view, which also asks for a corroboration in the history of science, despite the fact that Lakatos says the opposite. We do not hide the controversies, but these go in the Controversies section.

Dominic Mayers (talk) 01:58, 3 May 2020 (UTC)

Added a new heading for the two subsections

I finally realized that it was cleaner and simpler to add a new heading for the two subsections, which are more about problems within an approach to solve the problem of induction than about the problem of induction per se. The new heading is straightforward: "Problems of falsification". I changed the order of the sections. I moved the section Demarcation problem before, because it is more important. In fact, the section "Problems of falsification" could almost go after the definition of falsifiability. I also edited the first paragraph in the new section to better introduce it. I still think that we should expand on sophisticated falsificationism, but it will be easier to do that after we explained falsifiability. Dominic Mayers (talk) 18:33, 3 May 2020 (UTC)

I intend to add a subsection "Natural selection as a tautology" in the falsifiability#Examples of demarcation and applications section. Before I do that, I need to add another example of a falsifiable evolutionary hypothesis. These two additional examples about natural selection, one of a falsifiable statement and another of a non falsifiable statement, will together help to clarify some confusion on the subject. The subsections below are almost ready to be inserted in the article. Dominic Mayers (talk) 02:33, 22 May 2020 (UTC)

Minor point: not sure why you're citing a pdf of the 6th edition, when it's simple and more informative to link to specific pages in the DarwinOnline versions. For this particular point, On the Origin of Species#Struggle for existence, natural selection, and divergence has an inline citation to Darwin 1869, pp. 72, the first part of which links to the 5th edition in the references section. . . dave souza, talk 17:27, 20 May 2020 (UTC)

Thanks. Did not know it was possible to put a link in the page parameter of the harv templates. Dominic Mayers (talk) 18:15, 20 May 2020 (UTC)

Evolution

A famous example of a basic statement from J.B.S. Haldane is "[these are] fossil rabbits in the Precambrian era". This is a basic statement because it is possible to find a fossil rabbit and to determine that the date of a fossil is in the Precambrian era, even though it never happens that the date of a rabbit fossil is in the Precambrian era. Despite opinions to the contrary^[1], some times wrongly attributed to Popper^[A], this shows the scientific character of paleontology or the history of the evolution of life on Earth, because it contradicts the hypothesis in paleontology that all mammals existed in a much more recent era. Richard Dawkins adds that any other modern animal, such as a hippo, would suffice.^[2][3][4]

Another example is "In this industrial area, the relative fitness of the white-bodied peppered moth is high." Here "fitness" means "reproductive success over the next generation", which is the definition used as early as 1930 by Ronald Fisher^[B] and still the most widely accepted definition in modern biology (for example, see Cruzan 2018, p. 156, Muehlenbein 2010, p. 21, Ridley 2003). This is an example of a basic statement, because it is possible to separately determine the kind of environment, industrial vs natural, and the relative fitness of the white-bodied form (relative to the black-bodied form) in an area, even though it never happens that the white-bodied form has a high relative fitness in an industrial area. "In industrial areas, the black form of the peppered moth has higher relative fitness" is indeed a famous example of a falsifiable statement that illustrates the effect of natural selection.^[5]

Natural selection as a tautology

In the 5th and 6th editions of On the origin of species, following a suggestion of Alfred Russel Wallace, Darwin used "Survival of the fittest", an expression first coined by Herbert Spencer, as a synonym for "Natural Selection".^[C] Popper and others noted that, if one uses the most widely accepted definition of "fitness" in modern biology (see subsection § Evolution), namely reproductive success itself, the expression "survival of the fittest" is a tautology.^[D][E] In practice, as illustrated by the peppered moth example of section § Evolution, the questions asked are of the kind how specific traits affect the survival rate or fitness of a species when confronted by an environmental factor such as industrial pollution. Great Darwinist Ronald Fisher worked out mathematical theorems to help answer this kind of questions. But, for Popper and others, there is no (falsifiable) law of Natural Selection in this, because it only applies to some rare traits.^[F][G] Instead, for Popper, the work of Fisher and others on Natural Selection is part of an important metaphysical research program.

Notes for new examples

Abbreviated references

References

After reading more on the subject, I realized that there is more to say about non falsifiable statements like "All men are mortal". The section below has replaced the "All men are mortal" example of section "Simple examples of non-falsifiable statements".

Dominic Mayers (talk) 02:32, 22 May 2020 (UTC)

Grover Maxwell's useful metaphysical statements

Grover Maxwell called attention on an important class of metaphysical statements such as "All men are mortal". This is not falsifiable, because it does not matter how old is a man, maybe it will die next year. The law can be rewritten in the form "All men die at the age of N for some N." A key point is that, for all N, the law is falsifiable, but the difficulty is that we cannot run over all N. This is easily generalized to the form "All X are p for some property p of a given kind", where, for every property p of that kind, "All X are p" is falsifiable. When it is impossible to run over all p, the law is not falsifiable.

Popper's wrote that this kind of hypotheses are useful, because they suggest that we look for a property p such that the statement is highly corroborated. For example, "All men die before the age of 150" is highly corroborated. Thus "All men are mortal" is unfalsifiable, but it is nevertheless indirectly corroborated. Maxwell coined the term "corroboration without demarcation." Herbert Keuth added that, if we find that the statement is false for some p, the metaphysical law suggests that we look for another p where the statement is highly corroborated, say "All men die before the age of 250." Popper's idea is that, if no such a p exists (not even as a function of X), then the metaphysical law is not useful; it's not indirectly corroborated.

Melting point, beta decay and biological fitness

Another example from Maxwell is "all solids have a melting point." This is non falsifiable, because maybe the melting point will be reached at a higher temperature. The law is falsifiable and more useful if we specify an upper bound on melting points or a way to calculate this upper bound.

Another example from Maxwell is "all beta decays are accompanied with a neutrino emission from the same nucleus." This is also non falsifiable, because maybe the neutrino can be detected in a different manner. The law is falsifiable and much more useful from a scientific point of view, if the method to detect the neutrino is specified.

Another example, easily built from the pepper moth example, is "In all areas, the white vs black trait of the pepper moth affects its fitness". This is also not falsifiable, because may be the right environmental factor was not yet considered. When it is specified, namely, fitness in polluted industrial areas vs non polluted areas, then the law is falsifiable and it says what to look for.

Maxwell pointed out that most scientific laws are metaphysical statements of this kind, which need to be made more precise before they can be indirectly corroborated. In his paper "Corroboration without demarcation", Maxwell also considered the requirement for decisions in the detection of the neutrino, the determination of the melting point, which are issues discussed in section § Dogmatic falsificationism.

Dominic Mayers (talk) 16:02, 23 May 2020 (UTC)

The Chicago style suggests the following rules to place reference numbers in the text (emphasis mine):

Placement:

• Place a parenthetical note in your paper where you would put a reference number for a footnote or endnote: at the end of a quotation, sentence, or clause. You must include an in-text citation for every citation in your cumulative reference list at the end of your paper.
• Your parenthetical comes before any comma, period, or other punctuation mark when the quotation is run into the text.
• With a block quotation the in-text citation follows the terminal punctuation mark.

It seems confusing, because it's after the period when the reference is for the entire sentence, but before the comma when the reference is for what comes just before. Actually it is very logical. In particular, one should put the reference number before the period if the reference is not for the entire sentence.

Dominic Mayers (talk) 16:23, 23 May 2020 (UTC)

Wikipedia has its own Manual of Style! Wikipedia does not follow an external style manual (there are so many of them that editors would constantly be fighting over which manual to use). Ref tags are placed after commas and periods on Wikipedia. See: Wikipedia:Manual of Style § Punctuation and footnotes. Biogeographist (talk) 16:48, 23 May 2020 (UTC)

Also, the passage you quoted above is not about numbered footnotes, but about parenthetical references, which are different from numbered footnotes, even on Wikipedia. Biogeographist (talk) 17:49, 23 May 2020 (UTC)

@Biogeographist: Thank you.

I propose to remove the image of the red raven from the lead section. It does not add much to the swan picture and what is added does not complement any point my in the lead. The caption does illustrate an important point, which is that existential statements (over an infinite universal class) are not falsifiable, but again this is not a point made in the lead.

However, it brings out that the article should perhaps have a paragraph, if not a subsection, specifically about statements that contain an existential, including statements of the form ∀x ∃p p(x), where x runs over a universal class, say the class of all men, and p runs over a class of empirical properties on x, say properties of the form "x dies before the age of n", for some n.

The current version of the article avoids discussing existential. In particular, even though the section Falsifiability § Useful metaphysical statements is essentially about statements of the form that is mentioned above, it is not mentioned once in this subsection. In contrast, When Popper, Maxwell, etc. discuss these type of metaphysical statements, it is in terms of existential.

Anyway, for the time being, the point is simply that until we discuss existential statements, a picture to illustrate the concept is not useful. This picture might not be the best choice to do that anyway.

To say the truth, it is also that I don't think it is a nice photoshop work. The poor raven is blind. Maybe I should have mentioned this first.

Dominic Mayers (talk) 19:07, 24 May 2020 (UTC)

This is a draft of two new sub-sections related to the Popper-Lakatos controversy. Dominic Mayers (talk) 03:27, 29 May 2020 (UTC)

Methodless creativity versus inductive methodology

As briefly mentioned in section § Naive falsificationism, Lakatos accepted, as did Popper, that universal laws cannot be logically deduced, but he felt that, if the explanation can not be deductive, it must be inductive. He urged Popper explicitly to adopt some inductive principle^[A] and sets himself the task to find an inductive methodology. However, the methodology that he found did not offer any exact inductive rules. Indeed, in a response to Watkins and Musgrave, Lakatos acknowledged that the methodology depends on the good judgment of the scientists. Feyerabend wrote in "Against Method" that Lakatos' methodology of scientific research programmes is epistemological anarchism in disguise. In more recent work, he explains that there are rules, but whether or not to follow any of these rules is left to the judgment of the scientists.

Popper also offered a methodology with rules, but these rules are also not inductive rules, because they are not by themselves used to accept laws or establish their validity. They do that through the creativity or "good judgment" of the scientists only. For Popper, the required non deductive component of science never had to be an inductive methodology. He always viewed this component as a creative and irrational process, beyond the explanatory reach of any rational methodology, but yet used to decide which theories should be studied and applied, find good problems and guess useful conjectures.^[B] Quoting Einstein to support his view, Popper said that this renders obsolete the need for an inductive methodology or logical path to the laws. For Popper, no inductive methodology was ever proposed to satisfactorily explain science.

Ahistorical versus historiographical

Section § Methodless creativity versus inductive methodology says that both Lakatos's and Popper's methodology are not inductive. Yet Lakatos's methodology extended importantly Popper's methodology: it added a historiographical component to it. This allowed Lakatos to find corroborations for his methodology in the history of science. The basic units in his methodology, which can be abandoned or pursued, are research programmes. Research programmes can be degenerative or progressive and only degenerative research programmes must be abandoned at some point. For Lakatos, this is mostly corroborated by facts in history.

In contradistinction, Popper did not propose his methodology as a tool to reconstruct the history of science. Yet, some times, he did refer to history to corroborate his methodology. For example, he remarked that theories that were considered great successes were also the most likely to be falsified. Zahar's view was that, with regard to corroborations found in the history of science, there was only a difference of emphasis between Popper and Lakatos.

As an anecdotal example, in one of his articles Lakatos challenged Popper to show that his theory was falsifiable: he asked "Under what conditions would you give up your demarcation criterion?".^[1] Popper always viewed his philosophy as metaphysical, but in the context of Lakatos's challenge, which contained an emphasis on history, Popper replied "I shall give up my theory if Professor Lakatos succeeds in showing that Newton's theory is no more falsifiable by 'observable states of affairs' than is Freud's."^[2]

Dominic Mayers (talk) 18:39, 1 June 2020 (UTC)

Maybe we can add extra text that would be based on Goodman's work. Consider an universal inductive rule "given P1, P2, ..., Pn, we can infer C", which we know is invalid. Essentially, Goodman says that perhaps there is a corresponding valid universal rule "given L, P1, P2, ..., Pn, we can infer C", where L is a premise that can be established through investigation of P1, ..., Pn, C and anything else that is available. In Goodman's terminology, the Pi are instances of C and the premise L says that C is projectible. Goodman's new riddle of induction is then to find the method of investigation that can establish that a hypothesis C is projectible. Goodman proposes a method of investigation, but he says himself that it is "gross and tentative". For Goodman, there is no need to logically argue for rules of inference. The only criteria to judge rules of inference, Goodman says, is whether or not they work in practice. Therefore, the method of investigation makes use of history. Regarding the method, Goodman adds "We have no guarantees. The criterion for the legitimacy of projections cannot be truth that is as yet undetermined."

Of course, the plan is not to solve the problem of induction. In fact, it does not matter whether Goodman or Popper is right, because given that Goodman's criterion of legitimacy of induction is not yet fully established, it makes no difference. What Popper calls "creativity" becomes "explainable" by some unknown method that mimic induction, but in practice the methodology in which we apply falsifiability is the same. Anyway, the idea is to reduce this to a simple point that can be easily verified in the literature. The key point, I think, is that all rules in Popper's methodology are rules that scientists can interpret and eventually apply or not apply using their good judgment, which in some crucial cases can possibly be explained, if Goodman is correct, by some unknown method of induction.

Actually, Goodman uses "necessary" instead of "universal" and says that no rule is necessary. In other words, for Goodman, a conjectural decision based on past experiences is always needed to apply a rule. Goodman, of course, accepts that the rules of deduction are well established whereas the rules of induction are not. In fact, Goodman agrees that no rule of induction that respects his standard is known, so it's worst than not well established. Goodman's only proposes that we can find them and this is what he calls the new riddle of induction. In front of the same decision taken by scientists, say to pick a theory over another, Popper says that it was good judgment or intuition, whereas Goodman says that there was some method used, but we have not discovered it yet and this method would always be a conjecture anyway.

Dominic Mayers (talk) 20:10, 5 June 2020 (UTC)

Finally, I propose not to discuss Goodman in this article, at the least, not until it becomes clear how it enlighten important aspects of the current article. In any case, the question of Goodman new riddle of induction is better covered in The problem of induction. Dominic Mayers (talk) 19:34, 19 June 2020 (UTC)