Problem with science today and the war on reason

[Dr. Courtney]

I left out a key element which is that Popper's criterion hinges on falsifiability, not just testability. "X will evolve" is not a falsifiable claim.

One issue is that in evolution, the definition of evolution in specific predictions, as well as the specific nature of the predictions is harder to pin down. Usually, the claim of confirmation by prediction is made only AFTER the change is observed.

Using the definition of "change in allele frequencies over time" observation of evolution is trivial.

I think we can all agree that validated predictions of the future emergence of new species would be more impressive.

But even more interesting might be the ability to predict things like which species will evolve, which will move, and which will be extirpated in response to warming trends in specific regions.

If we really understood both evolution and global warming, shouldn't we be able to accurately predict which species have the capabilities to evolve fast enough to survive and which do not?

Aside from those mentioned by Dr. Courtney, a historical example of a concept in physics which had no predictive power at the time of its introduction yet developed into a core principle is Ockam/Buridan's theory of impetus.

Perhaps another example, if you'll accept it, is that of phlogiston. Of course, this theory was replaced rather than augmented as in the case of impetus, but the concept of of phlogiston (which is not too dissimilar to that of oxidation – backwards really) was widely accepted, had no predictive power and helped catalyze the development of more fruitful theories.

EDIT: Oh, and the theory of atoms progressed in a similar vain.

Great points. Scientists often miss the distinction between descriptive and predictive until the descriptive theory gets replaced with a better model with real predictive power.

Right, but in the case of medical research isn't there the risk of repeating studies on drugs (for example) that were shown to have no effect? It seems like having access to, and information about, null results could help streamline certain fields of research.

Of course. Null results should be published. But it seems to me that this requirement should be supported and enforced by the funding agencies rather than by external authorities. And in a business environment, why does one company who funded drug research care if another (competing) company wastes money on a dead end product?

Again, your examples aren't addressing the initial claim, that predictive power is somehow not a a defining element of science. The phlogiston hypothesis made predictions, thus it had predictive power, that just turned out to be wrong when experimentation tested those predictions.

I think you are confusing "predictive power" with falsifiability. Perhaps "predictive power" has not been defined as well as Popper defined "falsifiability." I do not see them as synonyms. For me, "predictive power" means able to reliably make CORRECT predictions of future events.

Falsifiability just means able to make predictions which are capable of being falsified in future experiments, whether or not they are.

So a theory that makes wrong predictions is falsifiable but has no predictive power.

A theory that makes right predictions is falsifiable and has predictive power.

 

[Evo]

One issue is that in evolution, the definition of evolution in specific predictions, as well as the specific nature of the predictions is harder to pin down. Usually, the claim of confirmation by prediction is made only AFTER the change is observed.

Using the definition of "change in allele frequencies over time" observation of evolution is trivial.

I think we can all agree that validated predictions of the future emergence of new species would be more impressive.

But even more interesting might be the ability to predict things like which species will evolve, which will move, and which will be extirpated in response to warming trends in specific regions.

If we really understood both evolution and global warming, shouldn't we be able to accurately predict which species have the capabilities to evolve fast enough to survive and which do not?

Since evolution doesn't fall into anything you mentioned, no. No, we don't understand either of them and so we will not discuss them.

If we really understood both evolution and global warming, shouldn't we be able to accurately predict which species have the capabilities to evolve fast enough to survive and which do not?

Why on Earth would you think that? I'm left without words. "We know how cancer forms, why can't we stop it?" And I could go on and on.

 

[Dr. Courtney]

Since evolution doesn't fall into anything you mentioned, no. No, we don't understand either of them and so we will not discuss them.

One issue is that in evolution, the definition of evolution in specific predictions, as well as the specific nature of the predictions is harder to pin down. Usually, the claim of confirmation by prediction is made only AFTER the change is observed.

Using the definition of "change in allele frequencies over time" observation of evolution is trivial.

I think we can all agree that validated predictions of the future emergence of new species would be more impressive.

But even more interesting might be the ability to predict things like which species will evolve, which will move, and which will be extirpated in response to warming trends in specific regions.

If we really understood both evolution and global warming, shouldn't we be able to accurately predict which species have the capabilities to evolve fast enough to survive and which do not?

Why on Earth would you think that? I'm left without words. "We know how cancer forms, why can't we stop it?" And I could go on and on.

We can only predict cancer in a probabilistic sense: a smoker has yy% chance of lung cancer, but a non-smoker has a zz% chance of lung cancer.

Our understanding will be much better when we can say, "Given your genome, you will (100%) or will not (0%) develop lung cancer if you smoke ZZ packs a day."

I have an identical twin with type 2 diabetes. That gives me a 95% chance of getting it myself. We know that in many cases, type 2 diabetes results from too much weight, too much unhealthy food, and too little exercise. But these are just risk factors that allow probabilistic risk assessments (at least for most folks without identical twins). I know with much more certainty than most that if I weigh over YY lbs, I will get type 2 diabetes. Won't we understand type 2 diabetes better when the same certainty of prediction is available for everyone?

And so it is with every theory (even your sacred cows), they will be better when they can make more definite and more reliable predictions.

And the fact that a scientific theory can be better in the future, means it is not as good as it can be now. For example, we've published work showing that cutthroat trout outcompete rainbow trout in lentic ecosystems. This allows the prediction that (in those systems), the genetic makeup will likely tend to shift in favor of cutthroat genetics, which is the opposite of what has shown to be happening in lotic ecosystems. In lotic ecosystems, rainbow trout outcompete and the genetics quickly shift away from cutthroat. At some point, improvements in understanding competitive and selection forces should allow more definite predictions in how allele frequencies change over time: which species will win and which will lose.

Theories with more specific predictions should always be favored over theories with only vague predictions or mere descriptions.

 

[brainpushups]

Again, your examples aren't addressing the initial claim, that predictive power is somehow not a a defining element of science.

Perhaps I didn't do a good job explaining myself at first. I'm not arguing that prediction is not an essential element of science. I was arguing that, in the early stages of scientific development the ideas are often (as Dr. Courtney mentioned) descriptive rather than predictive. Call it 'proto-science' if you'd like. I stand by my examples, but I don't want to waste space here arguing about it. Happy to do so elsewhere!

Getting back to the main point of the thread: it is impossible to tell which ideas will eventually become fruitful. The majority of them certainly do not. To quote myself:

As new conceptual frameworks emerge in science (even physics) they often don't have predictive power but can nonetheless develop into fruitful theories. I don't know when the right time for communication to the public about new conceptual schemes is appropriate. I see benefit here in generating excitement (perhaps just as the loathed-by-many popular accounts of modern physics do) but also danger based on what has been mentioned in this thread.

My main question was about the relationship between science reporting (especially of new ideas, but experimental confirmation could be part of this) and public perception. Taking the example of string theory - a widely written about subject in popular science - we have widespread discussion about a theory which, at the present time is arguably non-scientific (not because of prediction, but because of measurement). The idea generates interest in science while simultaneously (perhaps) harming public perception. So what is appropriate? As another historical side-note: Mach (who was a hard-core positivist) argued in the late 1800s that atomic theory was also non-scientific because there were, at the time, no measurement tools/techniques available to confirm or refute the hypothesis. Are we at the same stage here with string theory?

Also, getting back to education: the NGSS are attempting to focus student achievement on broader skills than just content knowledge. One thing that is not articulated well in any of the standards (to my recollection) is anything about how science progresses. If there was more focus on this then maybe the reporting of false findings or proto-science would be a non-issue because more people would understand how science evolves.

 

[Dr. Courtney]

Also, getting back to education: the NGSS are attempting to focus student achievement on broader skills than just content knowledge. One thing that is not articulated well in any of the standards (to my recollection) is anything about how science progresses. If there was more focus on this then maybe the reporting of false findings or proto-science would be a non-issue because more people would understand how science evolves.

One thing I love about the science portion of the ACT is that it requires very little content knowledge and focuses on the logical process of drawing inferences from given information in a scientific manner.

Students who have had strong lab courses and teachers who emphasized the scientific reasoning and thinking process tend to do much better on the science portion of the ACT than students from classes that focus overwhelmingly on content knowledge.

I like my students to read Bacon, Galileo, Wigner, etc. My preferred high school curriculum spends equal time on laboratory science as on content knowledge. Very expensive and labor intensive to implement though.

 

[brainpushups]

I'd be interested to look into this. Do you have any references that discuss it?

A Sourcebook in Medieval Science edited by Edward Grant. Page 275 contains excerpts from Buridan's treatise on the subject.

 

[brainpushups]

Of course. Null results should be published. But it seems to me that this requirement should be supported and enforced by the funding agencies rather than by external authorities. And in a business environment, why does one company who funded drug research care if another (competing) company wastes money on a dead end product?

But maybe running science like a business is part of the problem. I think that science would operate best if researchers could transcend capitalism. Knowledge for knowledge's sake! Hard to imagine getting back to that without scientists just being independently wealthy, but the profit motive of research is another one of the contributing factors to public distrust and dislike of science. I recall hearing years ago that certain drugs were being intentionally held back from the public because of patents. I don't know how true this is, but a quick google search uncovers a lot of popular articles on the subject.

 

[nikkkom]

What I see is that people doubt everything they hear or read - and this is a good thing.

Human civilization had HUGE problems when people were required to unquestioningly accept prevailing beliefs (religion, "the King has a divine right to rule over us", etc). It took many centuries to get rid of that.

Of course, in the population of billions of people, there will be crackpots which see a conspiracy and lies everywhere. (I just watched a flat earther youtube video which was discussing recent SpaceX launch and how "obviously fake" it is). Ignore the nutty ones (but accept that they will always exist); argue with those having doubts who nevertheless look rational.

 

[anorlunda]

Other factors:
- growing distrust of authority in general
- exposure to mass commercial media makes people skeptical of just about everything they hear
- binary vs spectral conditions: whether you accept creation or not is reasonably black and white, with climate change there are different positions: the climate isn't changing at all, the climate is changing, but humans aren't at fault, the climate is changing and humans are part of the problem, the climate is changing and carbon emission from human activities is the primary reason, etc.

I think you're on the right track. I attempt similar analysis, but my perspective is a bit different.

The most notorious clash between science and the public has to do with climate change. Climate science is one thing. Remedial actions (including possibly no action at all) is not science, it is a moral question having to do with values. Some scientists say publicly, "We predict climate change. Therefore you better do what we say regarding renewable energy or else you are an anti-science denier," I view that as scientific malpractice that misrepresents their opinions on human behavior and morality as being scientifically based. Even scientists whose field has nothing to do with climatology, are telling the public that only scientists should be allowed to have valid opinions about climate or climate remedies.

Another case. I know of a young girl studying MWI at Dartmouth. She plans to make MWI study her career. That means a lifetime living on public money, having no accountability or ever being required to produce a deliverable that advances the state of the art. When taxpayers object, they are told to shut up. As non-scientists they are not entitled to voice opinions. The public views science as yet another special interest group with their hands out demanding public money, but refusing to justify their demands except to fellow scientists. If you disagree, show me a scientific community that does not believe that more money, more power, more influence, more fame for themselves would not be a good thing.

I would not use the word authoritarian. I think "scientific priesthood" is a better description. Some in the public fear such a priesthood and imagine (rightly or wrongly) that they are being told by this priesthood "Hear, believe, and obey us on all matters that we deem science-related."

I also believe that we need a bit of realism in what we expect from public behavior. Public opinion is an extremely blunt instrument. There is no mechanism to accurately articulate the what and why of public opinion. When the public is angry at some group, their ire has no surgical precision, they simply lash out regarding anything and everything the target group does or says, often quite irrationally. By analogy, when lovers spat, it is common for them to say hurtful things that make no sense. Critics of the public focusing on such irrationality fail to address the root issues that cause the public anger in the first place.

 

[Dr. Courtney]

I would not use the word authoritarian. I think "scientific priesthood" is a better description. Some in the public fear such a priesthood and imagine (rightly or wrongly) that they are being told by this priesthood "Hear, believe, and obey us on all matters that we deem science-related."

Agreed. We have to reject attempts to construct and defend the "scientific priesthood" that is only accountable to other members of the "scientific priesthood." Some years ago, in arguing to retain Popper's falsifiability and Feynman's view experiment as the ultimate arbiter of science, a colleague and I wrote:

Defining science as “observationally constrained model building” is barely more specific than defining science as “what scientists do.” How far is this from defining sound science as “what scientists say” (with appropriate homage to peer review)? At this point, is science really a powerful, objective epistemology for exploring natural law, or have we merely replaced one set of authorities (the Catholic Church of the Middle Ages) with another (the scientists of the 21st century)?

See: https://arxiv.org/ftp/arxiv/papers/0812/0812.4932.pdf

 

[mheslep]

I think we can all agree that validated predictions of the future emergence of new species would be more impressive

Genetic changes are random and the long term natural process which selects them, leading to enitrely new species, is an interaction with a changing environment and other species which is non linear and chaotic. Predictions are not possible under those conditions. That is, sometimes KT impacts happen.

That said, I agree with the points above about predictive power association with the best science. If Evolution is weak in that regard, no apologies to Darwin.

 

[Dale]

Our understanding will be much better when we can say, "Given your genome, you will (100%) or will not (0%) develop lung cancer if you smoke ZZ packs a day."

If anybody ever says that then they are lying. That is not a reasonable expectation to set, we can't even say 100% that the sun will come up tomorrow.

There is nothing wrong with a probabilistic prediction. It is still predictive, and is more honest for inductive reasoning.

 

[Dale]

I think that there are three issues here, all three of which are contributors to a growing distrust of scientific institutions:

The first is an education problem. A lot of science instruction is teaching a collection Scientific Facts instead of teaching the scientific method and general inductive reasoning.

The second is the communication issue that has been hit on in the scicom discussion above.

The third is the current "replicability crisis". This is the one that I believe this generation of scientists most needs to address. However, I would take it in context. Scientific practice has been substantially refined over the years. Even a century ago it was uncommon to include error bars in any charts, today it is mandatory in many fields as a generation of scientists understood the importance. Similarly, statistical, reporting, and experimental methods addressing replicability are known within the community, and can be broadly adopted as their importance is recognized.

 

[Rive]

I would not use the word authoritarian. I think "scientific priesthood" is a better description. Some in the public fear such a priesthood and imagine (rightly or wrongly) that they are being told by this priesthood "Hear, believe, and obey us on all matters that we deem science-related."

I think it's a bit more difficult than that. If one wants to compare this to religious examples, then most scientists are not 'priests', but 'hermits' or 'saints' who study and develop the mystic knowledge. The priesthood was always a different class, which was expected to interpret and convey the mystic knowledge to the masses - but as can be seen in history this kind of relationship always ended in something else and so 'religion' (this case: science) always became a tool to keep up, promote or create authority.

Just look at the situation around GW and environmental protection...

 

[russ_watters]

If anybody ever says that then they are lying. That is not a reasonable expectation to set, we can't even say 100% that the sun will come up tomorrow.

There is nothing wrong with a probabilistic prediction. It is still predictive, and is more honest for inductive reasoning.

I agree with your point, but would clarify: "the sun will rise tomorrow" is not a probabilistic prediction, it is an exact calculation with a [small] potential experimental error in that the test might get interrupted by, say, Planet X careening through the solar system and splitting Earth in half. So it isn't a great example of a similar probability issue.

"Your odds of getting Type 2 diabetes are 95%" is a probabilistic prediction of exactly the same type as "This Pd103 atom has a 95% chance of decaying in 63 days." It is inherrently impossible for a probability to ever be 100% even excluding experimental error. That's just their nature -- they don't give exact answers. But if the question is framed differently, they can give predictions that are extraordinarily precise: "50% of the atoms in this 1g sample of Pd 103 will decay in 17 days".

So (agreeing with you) it is unfair to expect the prediction "95% of people with this gene sequence will get type 2 diabetes" to enable a high precision result for an individual in that population -- even setting aside confounding factors in the population, such as the individual subjects' ability to affect the outcome of the experiment!

@Dr. Courtney - the silver lining on your bad news that you are very likely to get type 2 diabetes is that your level of uncertainty is actually pretty low compared to others who get similar news: Angelina Jolie removed her breasts on more uncertain odds than that!

TL;DR - Probabilistic predictions can be extremely accurate, but they have to be used properly.

 

[Dr. Courtney]

@Dr. Courtney - the silver lining on your bad news that you are very likely to get type 2 diabetes is that your level of uncertainty is actually pretty low compared to others who get similar news: Angelina Jolie removed her breasts on more uncertain odds than that!

When someone assigns a 0% or 100% probability to an event, I give them the benefit of the doubt and do not assume dishonesty any more than I would if they assigned a 50% probability to an event. My tentative assumption is that they are rounding, hopefully using the rules of significant figures. This leads me to begin considering the uncertainty in the predicted probabilities.

But, being an optimist, I think there is a very good chance that behavioral modifications can put me in the 5% who do not get type 2 diabetes. Rather than take the science as an unavoidable sentence, I've used it to motivate effective action (which should be a big goal of most medical science). The error bars are bigger, but with exercise, dietary modifications, and weight loss, I can reduce my odds to well under 50%, at least within the first decade after a pre-diabetic diagnosis. And the way diabetes works, is an initial diagnosis at 60 rather than at 50 likely results in a much higher quality of life for the remaining years.

So, in some ways, it is about understanding the probabilities. In other ways, it is about reducing the uncertainties. Science is always better if scientists can better understand the probabilities and reduce the error bars. However, in a lot of fields, having an estimate of a probability with large error bars is great progress compared with the earlier state of affairs where there are no quantitative predictions or the quantitative predictions are really only guesses.

 

[anorlunda]

If I correctly interpret the subject of this thread is the public reputation of science. There are two themes running in the thread. Both are relevant to the thread's topic..

1. The difficulty of accurately conveying scientific knowledge to a public audience. Obviously, this applies only to the tiny subset of scientific knowledge that the public is interested in hearing about.
2. The public nonscientific behavior of some scientists acting individually or as a community. Science is but one of many communities vying for public trust and authority. In that sense, science competes with politicians, celebrities, journalists, religions, and others in a non-zero-sum-game of winning public trust.

The difference between the reputation of science and the reputation of scientists is a hair that I don't think is worth splitting.

 

[russ_watters]

When someone assigns a 0% or 100% probability to an event, I give them the benefit of the doubt and do not assume dishonesty any more than I would if they assigned a 50% probability to an event. My tentative assumption is that they are rounding, hopefully using the rules of significant figures. This leads me to begin considering the uncertainty in the predicted probabilities.

Ok...

But, being an optimist, I think there is a very good chance that behavioral modifications can put me in the 5% who do not get type 2 diabetes. Rather than take the science as an unavoidable sentence, I've used it to motivate effective action (which should be a big goal of most medical science). The error bars are bigger, but with exercise, dietary modifications, and weight loss, I can reduce my odds to well under 50%, at least within the first decade after a pre-diabetic diagnosis.

Agreed. But that's just a matter of how the doctor frames the discussion: indeed, that's probably the primary reason for having it!

The doctor is telling you the odds (if you don't attempt to change them) and telling you that you can change the odds. That isn't a flaw in the science, it is a separate/special power you are being given, to change the experiment!

So, in some ways, it is about understanding the probabilities. In other ways, it is about reducing the uncertainties. Science is always better if scientists can better understand the probabilities and reduce the error bars.

I don't want to make this pedantic, but it seems to me that the issue is more about understanding probabilities and therefore properly framing the question than about "error bars". Stating a half life of an isotope isn't a massively wide error bar (50% +/-50%?), it is extraordinarily precise probability. So:

However, in a lot of fields, having an estimate of a probability with large error bars is great progress compared with the earlier state of affairs where there are no quantitative predictions or the quantitative predictions are really only guesses.

I'd be curious as to what other issues/fields you would characterize probabilistic data/predictions as having wide "error bars" in a way similar to your genetic odds of diabetes... If you say "social sciences", then I think we really do disagree...

 

[Dale]

I agree with your point, but would clarify: "the sun will rise tomorrow" is not a probabilistic prediction, it is an exact calculation with a [small] potential experimental error in that the test might get interrupted by, say, Planet X careening through the solar system and splitting Earth in half. So it isn't a great example of a similar probability issue.

"Your odds of getting Type 2 diabetes are 95%" is a probabilistic prediction of exactly the same type as "This Pd103 atom has a 95% chance of decaying in 63 days." It is inherrently impossible for a probability to ever be 100% even excluding experimental error. That's just their nature -- they don't give exact answers

I think I understand what you are saying. You are drawing a distinction between cases where the uncertainty is due to lack of knowledge of the sources (experimental error) but the laws are exact compared to cases where the uncertainty is part of the laws themselves.

I am a "moderate" Bayesian, and the Bayesian approach strongly informs my thinking about the process of inductive reasoning. So, what I was getting at is that there is never a 100% posterior, regardless of the likelihood and the data (assuming an uncertain prior). This applies both to deterministic hypotheses and probabilistic hypotheses.

 

[anorlunda]

I am a "moderate" Bayesian

That would make an awesome t-shirt

 

[Greg Bernhardt]

That would make an awesome t-shirt

I can make that happen!

 

[Dr. Courtney]

I don't want to make this pedantic, but it seems to me that the issue is more about understanding probabilities and therefore properly framing the question than about "error bars". Stating a half life of an isotope isn't a massively wide error bar (50% +/-50%?), it is extraordinarily precise probability. So:

I'd be curious as to what other issues/fields you would characterize probabilistic data/predictions as having wide "error bars" in a way similar to your genetic odds of diabetes... If you say "social sciences", then I think we really do disagree...

No, I am thinking about most medical science and biomechanical risk of injury. Stuff like LD50 for toxins or overdoses. (Lethal Dose 50%, the dose that is expected to cause death 50% of the time.) The head acceleration that is expected to cause a given level of traumatic brain injury (mild, moderate, or severe) 50% of the time. The force on the heel that will break the tibia 50% of the time. The height of a fall onto a hip that will fracture the femur of an "average" 70 year old woman 50% of the time. The blast wave parameters that will cause lung injury (or brain injury) 50% of the time. The probability of incapacitating an enemy soldier with a torso hit of a given bullet and a given impact energy.

 

[russ_watters]

No, I am thinking about most medical science and biomechanical risk of injury. Stuff like...

Ok, then I think we do agree. It seems to me in medicine there are a lot of probability based statistics/predictions that don't have clear-cut cause-effect relationships or have a lot of factors that may influence the probability significantly, making it hard to isolate the impact of one in particular. I don't think your example is necessarily a great one, but I do agree the issue exists.

Those examples are ones where I'd say there are so many confounding factors with major influences that they often aren't useful...except as motivational talking points during a doctor's visit.

 

[OmCheeto]

I can make that happen!

I'll take 5: XL, 1 black, 1 white, and 3 pink.

 

[gleem]

The title of this thread is "Problem with science today and the war on reason " Does anybody have any data on the escalation of the implied distrust science in recent years. I have seen articles that say that about 20 years ago 70 - 80% of people trusted science. Is this "war" just a noisy refusal of a previously silent minority to accepts some facts from a few sciences that do not agree with their world view?

 

[jack action]

I don't think the war is on reason, it is about having "them" imposing a way of life on people. There are more and more laws constraining our daily behaviors and too often while invoking the name of science. It's really fun when the laws go our way, not so much when they don't. It's even worst if you're told you can't have a say because you are not an "expert".

That is what the war is about. Science is sadly caught in the middle.

 

[Dale]

I don't think the war is on reason, it is about having "them" imposing a way of life on people

I think that is a good point. There would probably be far fewer climate change objections if the climate change science were not being used politically to push through expensive policies. That is just a personal guess

 

[anorlunda]

I think that is a good point. There would probably be far fewer climate change objections if the climate change science were not being used politically to push through expensive policies.

We may be coming to a consensus. Such abuse of science for political purposes is an example of what I called scientific misbehavior. It tarnishes the reputation and credibility of the entire science community.

By the way, science will never reconcile with their public critics by using pejorative language like "war on reason." The people are sovereign. Problems are never solvable by insulting the sovereigns.

 

[gleem]

There would probably be far fewer climate change objections if the climate change science were not being used politically to push through expensive policies.

I'm missing something. How expensive might it get if the issues which science predicts are not addressed. Or are you saying there are no scientific issue to be addressed. Isn't it why we have a government so as to address problems when the citizenry is adversely affected.

We may be coming to a consensus. Such abuse of science for political purposes is an example of what I called scientific misbehavior.

So what abuse for political purposes can you identify out side of the current administration's denial of CC.

 

[anorlunda]

So what abuse for political purposes can you identify out side of the current administration's denial of CC.

The IPCC's perceived mission to persuade governments to adopt certain policies. Climate change science ends with predictions of future climates under a range of assumptions. Policy, any policy, is not science. Anyone advocating or opposing any policy when acting as a scientist and not just an ordinary citizen, is abusing science.