Falsifiability dethroned as the criterion of science?

I just read something that surprised me. The Wikipedia article on conspiracy theories started to note what’s unscientific about them, and of course pointed out that they’re not falsifiable. Then it said,

So I read the article on falsifiability and it didn’t say this explicitly. But in the section on its use as the “criterion of demarcation,” i.e. how to tell science apart from nonscience, it didn’t say that in so many words. But it implied as much, saying

as if to say that while it’s nice to have around, it isn’t the ultimate arbiter of science.

The last I’d heard, falsifiability was still in good standing among Dopers who use it to shoot down pseudoscience. It was the hero in an Arkansas evolution trial because the judge used it as the criterion for ruling that creation science isn’t really science. The Supreme Court has also established it as the criterion for whether evidence is scientific or not. I thought Dopers who keep their anti-pseudoscience tools honed would have heard of this development. I searched for “falsifiability” here and found lots of threads debating various specific pseudosciences, but no debates on the position of falsificationism itself in the philosophy of science. It seems to be assumed as a given that Dopers will uphold it as a necessary principle for fighting ignorance.

The articles also noted the arguments of critics who have been saying all along that Popper overestimated the value of his baby. Is it possible that these two Wiki articles simply reflect the views of the critics, and that in mainstream philosophy of science it continues to reign? Or is its day really over? What are the latest developments in epistemology anyway?

There is no “ultimate” anything in philosophy - any definition or criterion has possible criticisms, and one really has to plump for the least worst. I think falsification counts as least worst, by far, compared to its alternatives.

Well, definitions of English words are not static but are instead based on a consensus meaning of use. So “science” generally means “that which scientists engage in”. If everyone thought “science” was guessing how many angels dance on a pinhead, I’d have to use other words to describe collecting objective observational data and using them to predict and explain phenomena in a way which is at least theoretically falsifiable. But it would be the “scientists” who have changed their methods: the old method is still viable.

IMO, theoretical falsifiability is a necessary but not sufficient component of science. After all, the IgNobel prizes are awarded for results of studies that cannot or should not be replicated: I am perfectly content to let actual scientists determine the actual demarcation point at which a study is meaningful enough to be science. Granted, that is a political position somewhat, but even so, the persons doing the objectionable research are still studying their subject matter in a scientific manner even if it’s not science.

OTOH, some implications of some scientific theories are not falsifiable: for instance, if “miracles” (public events seemingly in contradiction to established scientific laws,) regularly occur but with no concrete evidence for a higher being, a proper scientific pursuit would be to record, analyze, and even postulate mechanisms by which “supernatural” beings perform these.

It would be fine for scientists to publicly opine that they think these miracles are firm evidence for a higher being, but they would not be speaking ex cathedra. In general, they have as much or more wherewithal to make philosophical analyses of their data and theories, because they are more intimately familiar with it than the general public, but it’s not necessarily science (IMO of course.)

I seem to agree with that, but this phrase is what puzzled me: “falsifiability is not now considered a tenable criterion…” It seems to say that falsifiability once was considered as such, but epistemologists have recently demoted it. I haven’t seen this idea anywhere else, which made me wonder if the author was just a random crackpot whose Wiki assertion hasn’t been caught and corrected yet, because anyone can publish any nonsense on Wiki before someone comes along and changes it. Or is this really true? It made me realize I haven’t been keeping up with the latest developments in epistemology. Ludovic, the article on falsifiability agrees with you, explicitly calling it a “political position.” Unfortunately, the introduction of politics into the scientific method just sounds all wrong. Interests of a political nature are fundamentally at odds with the search for objective facts.

“Objective facts” is an oxymoron in a world percieved by subjects. Falsifiability is a good criterion if the target of inquiry is assumed to be invariant, or if the domain of inquiry can be encompassed.

Don’t forget that Wikipedia can be edited by just about anyone, and if a particular author has a strong bias, they can really mess up an article. There are certainly critiques of falsifiability as the ultimate criterion for what is science, but as far as I know, they aren’t widely accepted.

Johanna, the Wikipedia article agreed with me in that falsifiability shouldn’t be a bright firm line of demarcation between science and not-science, but it says that

Which I agree with. However, while perhaps scientists are better equipped to deal with it than non-scientists, non-falsifiable hypotheses are still non-scientific, IMO. It’s just that I don’t think we need to elevate every falsifiable area of study to the level of science. I don’t necessarily agree with the political decision to exclude some areas (such as some engineering or social studies,) but it is perhaps inevitable given human nature and the English language.

Science is the attempt to systematically develop ever more accurate models of how things work. Such a process of model improvement require constant doubt as to whether or not the current model has it quite right. Falsifiabilty is thus desired but not always required; a model may be replaced not because it was falsified, but because another model fits the data better, or fits more data more comprehensively.

There are certain highly technical critiques of falsifiability about which I’m not fully competent to expound, though the wiki article is a good start. There are some good book length critiques of Popper if you’re really interested. However, for the purposes of a message board, a rough and ready idea of falsifiability is a useful tool when confronting intelligent design or new-age crackpottery.

The standard Popperian view of falsifiability falls down on two grounds.

Firstly and most simply, it is an absolutist position. A single negative result is sufficient to nullify a hypothesis. Much of modern science relies of statistical measures of certainty. For example, in a modern drug trial, you might have 33/50 patients healed with a drug and 12/50 patients healed with the placebo. From these numbers, you can then derive a confidence that your hypothesis (namely, the drug you are testing is effective against the diesese) is correct. Under such a double blind study, there is no set of results that can render a hypothesis completely false. Even if you had 0/50 healed for your drug and 50/50 healed for the placebo, there is a trivial but non-zero chance that your drug might actually work. Thus, drug trials are not strictly falsifiable but are undoubtable modern science.

Secondly, Falsifiability only works for experiment based sciences, not evidence based ones. In experiment based sciences, you are allowed to some degree to control your variables and to run an experiment as many times as you want. In evidence based ones, artificial experiments are not possible and you must decipher the results of natural experiements. The example I like to use to explain this comes from Jared Diamond’s book, Collapse. In this, he posits that there is a correlation between the size of polynesian islands (among other things) and how deforested they become. Now, this is clearly a scientific hypothesis. The problem is, there are only a finite amount of islands (81 to be precise). The only way to falsify it would be to discover an 82nd island which showed an opposite trend but this is clearly absurd. When asking some questions, the sample size is big enough such that you can essentially ignore the lack of falsifiabilty. ie: dinosaur fossils are being discovered at a large enough rate that they essentially serve the same role as experiments in terms of falsifiability. However, in other cases, only a tiny handful of examples exist or the examples differ so much from each other that essentially no hypothesis can be falsified. Hypothesises about a particular species for which only a few fossils exist. Or about whether a certain greenhouse gas is a significant cause of global warming. Or what factors affected planetary formation. In all of these cases and many more, the evidence that exists is not enough to conclusively falsify any of some mutually exclusive predictions and there is no mechanism for gathering more yet they are all undoubtably science.

Falsifaible science is merely a subset of science in general. Falsifiability hasn’t fallen by the wayside as a criteria, merely superseded.

I like Dr. Carlo Lastrucci’s definition in his book, The Scientific Approach

It seems to work for me.

That’s falsification, not falsifiability. We’re not talking about “correct” theories here, but scientific ones. (Correct in quotes to indicate that no theory can be proven.) The Steady State theory, for instance, is wrong but scientific.

Right, one thing I gathered from reading those articles is that falsifiability is not an experimental principle, but a logical one. The point is not that no counterexamples among Polynesian islands have been discovered, but the point is that Diamond’s theory is so formed that it’s conceivable a counterexample could falsify it. Just supposing.

I agree what you’re saying, but I have to quibble with your example. Diamond’s observations about Polynesian islands are not hypotheses because the entire population of Polynesian islands is available for study and he’s just crunching numbers. Now if he were trying to explain the observations, anything he came up with would be a hypothesis (and most likely falsifiable, if not in the strict Popperian sense).

I’m not strictly familiar with the writings of Popper, but I’d be very surprised if he didn’t make a distinction between theories that are falsifiable in principle and theories that have sufficient evidence to make a decision. In all the examples you cite, there is at least one theoretical piece of evidence that could overturn the theory.

The OP’s article talks about “statements”, but I think the key is to focus on hypotheses. While there might be some non-falsifiable statements that are of use to science, of what use would a non-falsifiable hypothesis be? If we define “science” as the scientific method, then at some point we have to verify a hypothesis. If that hypothesis cannot be proven wrong, then how does is fit into the scientific method? And by “cannot” I mean even theoretically, not just practically.

As for ID, it’s not inherently unscientific as long as the designer is not postulated to be a supernatual being. I highly encourage the ID folks to restrict their hypotheses to the natural world and to “do science” that can be published in peer review journals. So far, I haven’t seen anything that qualifies.

There is another distinction we need to make: some hypotheses are not falsifiable, even in principle, and are thus not scientific, and some are, but have advocates who refuse to accept evidence of falsification.

An honest creationism, where God can be invoked to explain away all the implausibilities, is an example of the first. “Scientific” creationism, which halfway keeps god out of it, and tries to claim Biblical events happened through supposed evidence, is an instance of the second.

There is a continuum. While scientists who refuse to face the evidence beome laughingstocks (like Hoyle or Pons and Fleishman) few are willing to give up a hypothesis based on one piece of contradictory evidence. That’s why there are so many interesting arguments about things.

I think you the OP is just hearing the echos of logical positivism’s dying gurgles.

The difference is between methodology and philosophy. There’s no way to prove that falsifiability is the way to determine “the truth.” But methodologically, in practice, you just can’t do science without some claims that can be shown to be either true or false by evidence. Go ahead and try! The proof is in the pudding.

There’s an ongoing, increasingly rancorous debate presently in theoretical physics circles about the importance of falsifiability. It would appear “science” these days is pretty much what some scientists want to define it to be, even if it starts to look like an abstruse exercise in modal logic.

Voyager, I acknowledge your point and retract, especially given John’s points.

Apos, as an obnoxious aside, the real saying is: “the proof of the pudding is in the eating.” That way makes some sense. What, did some mathematical genius cook his homework?

Is that your assessment of some of today’s theoretical physics? How so?