Take a look-see at this science pic:
This is the output from some model, the nature of which is not especially interesting, and the accompanying observations. In other words, predictions and the resulting values.
See the red arrow? I picked this point because it’s easy to see; there is nothing else special about it.
The blue diamond the arrow highlights was an instance at which the model said, “The value will be 240” (or whatever). But the observation was 70 (or so).
Very well: the model said 240 and Nature said 70. The model is therefore falsified. The model said a thing would happen. It did not happen. The model has said a false thing. The model was wrong: it was not right. It is falsified, so throw the model out. Right?
We have discussed the philosophy of falsifiability in detail before. Today, I want to speak of the sociology of it. Do not mix them up; however, the latter relies, in part, on the former, so we can’t entirely separate them.
Next cast your eye to the other blue diamonds. Sometimes the model said what would happen, or near enough, and sometimes the model did not. The model has been falsified not only once, but many times: indeed, each time the blue diamond does not fall on the red 1-to-1 line the model has been falsified again.
Now to the sociol—-
“—Wait, Briggs. Nobody takes the predicted values perfectly serious.”
What do you mean?
“I mean that when the model says ‘240,’ everybody takes that to mean ‘240 plus or minus a little.'”
So that if the observation goes outside that “little”, the model is then falsified?
“Not quite. Because most other predictions, the whole of those blue diamonds, might be close enough.”
How many have to be “close enough”?
Sounds pretty vague and subjective, this falsifiability. It only works when you say it does, based on rules that can’t be communicated quantitatively.
It also sounds like you mean that, in math, we can work out that each model prediction is “ackshually” a probability, like the mean of a normal distribution, or whatever. But that implies adding another model on the original model, a probability model (if the model itself doesn’t have this feature). Then each prediction is given a non-zero probability of happening. That 240 might have been given a probability of, say, 10%. It turned out to be 70, which the model gave a probability of, say, 0.1% to.
In other words, the model said the 70 was possible: 70 happened. The model is not falsified. So we can keep it.
But then, no observation can ever contradict the model, not formally. Because the probability model will never say any point is impossible.
Formally, falsifiability is useless.
Now let’s turn it around and look at a model that I say has had enough evidence of its failure, one so bad that no earnest person can believe it. A model that has had a century’s worth of dismal performance, which has never worked anywhere in practice. Yet which is still believed—and even cherished!
I call this model Masks Stop The Spread Of Respiratory Diseases, or Masks for short. Hard to find an Expert or ruler who isn’t a True Believer in Masks. Here’s a fun picture:
Deaths in New Zealand are rising again and near their record high, despite 95% of everyone over 12 in the country being fully vaccinated and a mask mandate for the past 9 months
Somehow no one in the media seems interested in asking how that could be possible pic.twitter.com/xuPrOH6txc
— Ian Miller (@ianmSC) May 16, 2022
This fellow has dozens upon dozens of pictures like this; others contrasting areas with mask mandates with those that don’t. Usually, the areas without mandates do better, in the sense of lower deaths or infections (mistakenly called “cases”). I haven’t seen any pics where maskless areas obviously do worse.
I have dozens of old studies, including the time when surgeons stopped wearing masks in a busy hospital for six weeks and there was no change in infection rates, all starting from a century ago, all showing the uselessness of Masks. Further, this information is easy to find, and easy to read, especially with those who have any training in medicine or statistics, and so should not be a mystery to anybody.
As far as falsifying a model goes, Masks has been as falsified as it can get.
Yet I ask you: is Masks still believed? Do elites, Experts and rulers still laud and praise it?
And why? Ah, that’s the real answer. Because any model that’s useful is a good model, falsifiability be damned. Both models detailed today are useful—to certain people. The trick is understanding that the model doesn’t have to be useful for the things of which it speaks. It merely need be a device to point to to justify an action one desires.
For the first model, it’s agriculture. For the second, it’s health. Both models exist for governments to say “We are doing something. And that something is The Science.” The simple models become meta models, part of a larger scheme. It’s those meta models that need examination.
The only conclusion is: however valuable it is in rigorous formal logic and mathematics, falsifiability is useless in science. Don’t bother with it.
Buy my new book and learn to argue against the regime: Everything You Believe Is Wrong.
Subscribe or donate to support this site and its wholly independent host using credit card or PayPal click here; Or go to PayPal directly. For Zelle, use my email.
When debating whether the concept of truth has a place in scientific philosophy, I like to ask my colleagues (who largely say it does not) whether, when they say that X is falsified, it is a true statement that X is falsified.
Of course one can dodge out of the way of that by insisting that X is only every falsified at some level of probability, but then one night ask if the statement ‘X is at six-sigma tension with the model’ is a true statement. Continuing on from there one rapidly reaches pure epistemic nihilism, at which point one reasonably asks, why are we even bothering to do science then?
Which is a question more scientists would do well to ask, to be honest.
“…falsifiability is useless in science.”
“The only conclusion is: however valuable it is in rigorous formal logic and mathematics, falsifiability is useless in science. Don’t bother with it.” – true, but complicated: how do we get to know things, apart from our senses and what can be ascertained immediately by our intellects?
As an institutionally trained economist, my instincts say that we are ALL thinking way past the marginal limits of any reasonable return on cognitive effort. If something is truly falsifiable the government (via our chosen representatives) will notice it straight away and courteously stop us from believing in what is against our own best interests.
My eventual conclusion after pondering the question for over thirty years (including a longish period of belief in Popperian ‘falsifiability’) was that there is not, nor ever has been, any special ‘method’ to science – beyond that scientists *must* seek and speak truth regarding their area of focus; and this must be their deep and sincere driving motivation.
With that, science sometimes happens if some people of sufficient ability put-in sufficient effort – but without truth as motivator we just get a variant of top-down-controlled bureaucracy: as now.
The reason why George Box said that all models are wrong is that no model includes all the factors in the real world. It is never the case that Hypothesis?Conclusion. It is always H1 and H2 and H3 etc. ? C. Therefore, when we see not-C, we don’t necessarily know which H was falsified. All scientific laws, as in every one of them, is ceterus paribus. “all else being equal.” Try dropping a cannonball and a dollar bill from a tower and the heavier one hits the ground first. That’s because something other than gravity affects the motion of the dollar bill but has no measurable effect on the cannonball. Let’s call it “windage.” (Now crumple the dollar bill into a ball and drop it and it is clear something other than mass is involved, as well.) Thus, a scientific law is valid only within the restricted domain within which it was measured.
I’ve always thought when people say “its not falsifiable” they meant “its circular logic” and always thought it would be bwtter to just to say it that way. The term “falsifiable” is confusing. It itself is meant to deceive and make people think that to be true something must be proven false. Lol. Better to just say “that’s circular logic.”
Is that the baby going down the drain with the bathwater?
Sure seems like the issue here is not with falsifiability.
Rather, it seems that the issue is with human nature, or “Sociology.”
If humans ignore the results of their experiments, that does not mean the results are useless.
My backup warning system beeps to tell me there’s a pole behind me, but I ignore it and keep backing, right into the pole. That doesn’t mean that the warning system, or the beeping warning, are useless. It means that I’m a fallible human, subject to the vicissitudes of human nature, psychology, and sociology.
It would seem that the logical conclusion to this conundrum is: “Science,” as practiced is an extension of politics. Care must be taken to disentangle the two. And: Ignore falsified hypotheses at your own peril.
Summary: Truth and reality have little or nothing to do with politics.
Science, since it involves humans doing human things, is a branch of politics.
Therefore, truth and reality have little or nothing to do with science. Or rather, ¡Science!
Perhaps I’m misguided, but I got the impression that Popper introduced the idea of ‘falsifiability’ in order to encourage a narrower scope of experiment or investigation – in a sense to correctly identify or define the elements of a particular problem – in such a way that each can be measured and tested.
True that this technique may not be suited to encompass the real world with respect to clinical, evidence-based and population study. The real world is just too complex.
But it does suggest that we break a problem down into small pieces, each one with it’s own, potentially measurable, uncertainty. Eventually these small steps may reach a larger goal that becomes useful; even if that result proves fruitless, at least some knowledge has been gained.
Isn’t this the entire purpose of scientific endeavour? In this respect I believe Popper’s ideas carry some importance.
Of course, Politics and Science do not mix. One always corrupts the other. The unfortunate situation we see today in the West.
The problem is that in most branches of modern science there is never as clear of failure state as backing into a pole. Take the kind of example that Briggs used in the article of a line of best fit being applied to data. If we said that such a model was “falsified” if even a single point was off the line, then every single such model used for real world data would be falsified. Some fudge factor is necessary, but what is it? It would almost certainly depend on the application. But that means that the fudge factor itself is a hypothesis which, in the world of falsifiability, only can be true to the extent that it can be falsified.
So for example suppose you set up a line of best fit and the data is only close to it, but not on it. I say that your model is falsified because the data is not within epsilon of the line, but you say that in this context the appropriate fudge factor is two epsilon, and that error bound does contain all the data. How can we determine which one of us is right? This isn’t a question of dishonesty; we could both be honestly viewing the data and it really could be that depending on the application a fudge factor of either one or two epsilon is appropriate. But we cannot know which is appropriate just from the data itself, at least not without knowing the “correct” model line, which is what we were trying to determine in the first place!
In order to make actual progress we need to induct to physical laws with a causal explanation. So for example the idea of gravitation did not come about based only on trying various ways to fit data and falsifying things until only that remained. It came by intuiting the law “gravitational force is dependent only on the distance between objects and their masses.”
Sorry to be dense, but isn’t it the other way around in your first example? The blue diamond is the observation and the red line is the prediction right?
Nope. The blue diamond is both the prediction and obs. The red line is just for the eye, a 1-to-1 line, on which all perfect predictions would lie.
I misread it at first too. The x-axis of the chart is observed value, the y-axis is predicted value.
Nope: Falsifiability is not just interesting, but critical to the advancement of science.
1 – a single counter example would falsifiy the riemann hypothesis (and about half of all doctoral work done in number theory). Similarly, there is a great deal of interesting work going on in physics because transforms whose errors (mostly info losses) lay below measurement error until fairly recently are now showing up as questionable because the predictions produced by these (implicit) models are being falsified by newer, more accurate, measurements. As a result the entire general model may be about to undergo significant revision.
2 – in your article you confuse science with application. People who, for example, “model” overseas barley sales by regressing daily sales data against time produce nice straight trend lines with ridiculously many outliers. This is application, not science -and these outliers do not falsify the model because there is no causal assumption – the person doing this merely wants to know whether sales are trending up or down and by very roughly how much per year or quarter and does not think that Mondays “cause” sales, he merely notes that more barley is bought on Mondays than any other day of the week and so thinks next Monday’s sales will be higher than Friday’s.
3 – where the people doing this kind of thing confuse science with application the conclusion you draw is often correct, but so what? these people have no idea what they’re doing besides advocating for some belief or policy: their implicit model is not related to the data, it’s related to how they think their views are perceived by their audiences.
No. He emphasized modus tollens because positivism was not capable of reaching sure conclusions. No number of observations can conclude to a universal Truth, but a contrary observation can lead to a certain Falsity. That is, you can disprove an hypothesis more surely than you can prove it.
But you must take care about the nature of the falsification. Copernican heliocentrism was not falsified by the lack of observable stellar parallax, but the distance to the stars was. (Though it was another two centuries before telescopes became fine enough to see the parallax.) The dilution of a mutation did not falsify natural selection, it simply showed that inheritance could not be analog (though it fell to Mendel to show it was digital). Permanent magnets did not falsify Maxwell’s electromagnetism, as some thought; it falsified the notion that there was no such a thing as an electron. The reason for this difficulty is that there are always multiple causes affecting the effect.It’s not just heliocentrism; it’s heliocentrism plus stellar distances plus other stuff. Hence, it is not always clear which of the factors has been falsified. (cf. the “E/M Drive.)
To follow YOS’s comment that “positivism was not capable of reaching sure conclusions”. Popper and others never saw that this was circular. If is is true that positivism is not capable of reaching truths, then etc.
Also, every thing falsified is its logical opposite truthified.
Even when you include a probability distribution, falsifiability still has a place, because, as you stated, the model describes the mean of the normal distribution. If the mean described by the model is deviated from by the distribution of observations, by some margin of error, then the model is falsified.
Yes, a wide range of observations ARE possible, but that is consistent with quantum mechanics already. That doesn’t falsify anything.
Dear Dr. Briggs, Are you perhaps a fan of Paul (Anything Goes) Feyerabend? I am, for the sheer fun of his apostasies.
Since I have a Ph. D. in engineering, and taught engineering for years a a major research university, I naturally had an interest in what philosophers of science had to say about what I was doing. Of course, Popper rules most minds in science and engineering, except in quantum mechanics, where he is anathema. However, I never found any philosopher who was relevant to my work.
And, of course, we always had to do p-values and confidence intervals in those days. I was never able to convince any graduate student that an elliptical, or even circular, scatter of data points was meaningless. After all, he had done all the hard work in the lab, and that must count for something. Also, he had a regression line and a correlation coefficient.
Nowadays, I regularly read Stephen Hsu’s blog “Information Processing,” and he regularly posts scatter plots that have a great deal of dispersion, but he also has fitted curves. Most of his plots are even worse than the first you cited.
By the way, are you surprised that the model did not calculate points for the excluded region, >360?
Whatever. The usual parody of Max Planck’s observation about scientific theories, that “science progresses one funereal and a time,” is apropos.
<blockquote?I never found any philosopher who was relevant to my work.
Have you tried Nancy Cartwright?