Ranking academic disciplines from science to pseudo-science
When I attack some specific study in say, sociology, on the obvious grounds (not defining its terms correctly, not knowing what it’s measuring, etc.), I am inevitably met with the same response: it isn’t physics! I.e. it isn’t physics, so I just shouldn’t expect it to be as sensible as a physics paper would.
But this is just defining away the problem: my expectation has no bearing on whether the paper makes any sense or not, it’s still rubbish. It’s like a religious person saying “this is outside the realm of science/logic” or that “there are some things that you just cannot test” – rejecting logic/statistics does not make what you say more intelligent.
Nonetheless, such an objection raises an important point: the problem is not just with the individual study – the problem is with the entire field. There are entire fields that simply remain unlit by the scientific flame to the present day, to which no one has built any signal connectors, despite a third of a millennium having passed since the publication of the Principia.
And since the lesson of mathematics, and the lesson of human thought in general, is to abstract when you can, I will classify the several “types” of disciplines, and the kind of unscientific notions present in them, so we can be more precise than simply call something “pseudo-science”.
Note 1: Note that these are sociological descriptors, not fundamental philosophical ones. These descriptions and classifications have nothing to do with the content of the fields themselves, but how they are studied by humans. Fields can – and have, throughout history – progress through these charts.
Note 2: But this is also not necessarily commentary on mainstream or consensus views. When competing schools exist in an area, I will choose the “most scientific” school for my classification and commentary.
Note 3: Although theory-based sciences are often called “applied mathematics” and model-based sciences are often called “applied statistics”, this is not the correct way to think about them. Statistics is just as fundamentally important to the actual use of any of the theory-based sciences – the distinction is that the “Bayesian prior” in theory-based sciences is much more solid, while in model-based sciences, it is essentially based on various cognitive biases, which may or may not align with what is correct. And these are not “applications” of anything in the same sense that “applied sciences” are applications of science.
1.
Tier-I: Theory-based sciences (“Applied mathematics”)
- *Physics (including chemistry, geometric optics etc.) *
- Economics
- Computer science
- Control theory, queuing theory, etc.
These sciences are fully based on a (usually elegant) fundamental, mathematical theory with a clearly defined domain of applicability which defines the scope of the discipline, and this theory can in principle make every possible prediction relevant to this domain of applicability. Terms and definitions are grounded in what makes the most sense on a very fundamental, natural level, rather than for any social or other superficial purposes.
Note that when I say general theory, I’m OK with even things that physicists would call “models”. The standard model may not be elegant enough to fit a physicist’s definition of a theory, but it’s on a totally different level compared to the “models” in the next category of disciplines.
2.
Tier-II: Model-based sciences (“Applied statistics”)
- Finance
- Econometrics
- Climate science
- Epidemiology
- Traffic theory
- Cognitive psychology
- Corporate research
- Operations research
These sciences do not have a general theory, they essentially run on rather specific, rather dubious data mining runs. Often, the metrics they study and their definitions of terms seem completely arbitrary on natural/fundamental grounds (e.g. IQ tests as the definition of intelligence, definition of “mental disorder”), and people are just supposed to try to come up with better definitions with the ones we already have, based on what the purpose of the definition is (e.g. as a predictor for something), in terms of the efficiency of the definition towards that goal (e.g. how good the causations are).
These metrics often come into conflict with people even further down below, who instead of critiquing them for not being scientific or theoretical enough, critique the metrics for being biased in various ways, because said critics have a very social, rather than natural understanding of the world and regard definitions as “constructs”, “social constructs” rather than as “definitions”. Needless to say, accepting said critiques would be a much more serious regression.
Their purpose should not be seen as advancing human knowledge in a general, abstract way, as the methods of these science are not truly reliable on a fundamental level – but they can (if done “well”) be reliable enough to be expected to deliver some form of profit to whomsoever it may concern.
3.
Tier-III: Encyclopedic sciences
- Biology (including genetics, neuroscience, anatomy, evolution)
- Geology and astronomy
- Technology (e.g. engineering, medicine)
These sciences do not have a general theory, but instead have an encyclopedic repository of facts. Although it may seem impossible to come up with a general theory of any of these fields that would predict all these facts: remember, chemistry was at one point also in this category.
Back when it was known as alchemy, before it was promoted to “modeling-based sciences” by Mendeleev and then finally to “Newtonian sciences” during the 20th century, chemistry was also just a repository of facts, a database of substances, etc.
Engineering and medicine – although not really sciences, but “applied science” – also fit the bill. The kind of thinking that goes into the encyclopedic sciences fields is exactly like that which goes into the applied sciences, because a biologist’s understanding of the human body is so incredibly basic he could as well be studying a piece of technology someone else built.
Even though this is ranked below “modeling-based sciences”, the facts established in these fields are often far more objectively clear than those in the modeling-based sciences. This is because these fields attempt less, they don’t claim to explain much beyond surface-level facts.
4.
Tier-IV: Phlogiston sciences
- Social psychology
- Parts of economics
- Parts of climate science
- Futurism
- Ancient science
- Pop psychology
- *Appeal to authority and tradition *
These disciplines are based on incredibly superficial and imprecise “explanations” and pseudo-theories that don’t actually explain anything.
While Freud is the classic example, the phenomenon is far more general in social psychology. You have all these “phenomena” that psychologists define: “projection”, “the Dunning-Kruger effect”, “Stockholm syndrome”, etc. But what does it even mean to define this phenomenon? That somebody was observed to have it? If I came up with a new psychological phenomenon called “optipessimisim”, and defined it as someone alternating between an optimistic and a pessimistic mindset, would that be accepted as a psychological phenomenon? What is the criterion for notability? Prevalence?
If you had an actual neurobiological explanation of what’s going on, I would understand: otherwise, it’s just your empathy replacing scientific methods. More generally (than just social psychology), it’s some very superficial intuition replacing scientific methods.
But the explanations your empathy comes up with are nothing better than the phlogiston theory of heat, they don’t actually explain or predict anything, it’s just a name you’ve given to something.
(Yes, there were “tests that disproved the phlogiston”, just like there were tests that disproved the ether. But these tests really disproved a specific model of the phlogiston, and people could’ve simply reduced the predictive power of the theory until it was completely vacuous. They didn’t, because they were physicists, and had the right mindset.)
Something similar applies to some claims in economics, such as the Keynesian multiplier. Maybe it exists, maybe it doesn’t, but the justifications you hear of it in the literature are just vacuous, made-up drivel that doesn’t really explain anything, just appeal to your superficial intuitions. And the same goes for parts of climate science, when you hear people talking about stuff like “polar vortex realignment” and using it to predict various natural disasters, etc. – these are not predictions they could make in advance, before seeing the data, from a theoretic foundation, they are fake explanations.
A quote from Eliezer Yudkowsky’s Harry Potter and the Methods of Rationality (HPMOR) is relevant:
“There’s a tale I once heard about some students who came into a physics class, and the teacher showed them a large metal plate near a fire. She ordered them to feel the metal plate, and they felt that the metal nearer the fire was cooler, and the metal further away was warmer. And she said, write down your guess for why this happens. So some students wrote down ‘because of how the metal conducts heat’, and some students wrote down ‘because of how the air moves’, and no one said ‘this just seems impossible’, and the real answer was that before the students came into the room, the teacher turned the plate around.”
“Interesting,” said Professor Quirrell. “That does sound similar. Is there a moral?”
“That your strength as a rationalist is your ability to be more confused by fiction than by reality,” said Harry. “If you’re equally good at explaining any outcome, you have zero knowledge. The students thought they could use words like ‘because of heat conduction’ to explain anything, even a metal plate being cooler on the side nearer the fire. So they didn’t notice how confused they were, and that meant they couldn’t be more confused by falsehood than by truth.”
It’s kind of like Sherlock Holmes’s “deductions”. Post-hoc explanations, rationalizations that don’t actually mean or predict anything precise.
Ancient science is mostly like this, and so are people who believe in “alternate medicines”, etc. today.
It’s not that these methods are necessarily bad ones – like in “Model-based sciences”, the methods of these disciplines could just work, if done correctly. I certainly think it could work for futurism if you have the right intuitions. But you’re not really following a fundamentally good method, and the certainty of your answers is quite low. It is important to acknowledge this.
Also on the same level we have beliefs on basis of endorsements by experts or traditions. These are weak indicators of truth, superficial intuitions, and the low certainty associated with said beliefs should be acknowledged.
5.
Tier-V: Cargo cult sciences
- Sociology (and the various studies that are its subfields)
- Humanities
- *Seerism (e.g. astrology, palmistry, numerology, prophecy, fortune-telling) *
These have all the faults of the phlogiston sciences but are much worse, because not only is the theory ill-defined nonsense, even their predictions are ill-defined nonsense based on narratives instead of fact.
In the examples shown here, the said “theory” is Marxian social theory, the shining example of pseudo-science. In short: the theory is built – from axiom to prediction – on imprecise terms such as “class”, “oppression” and the vary Marxist economic system. No actual reasoning is given for the predictions, and whatever reasoning is given is based on emotional conspiracy-theory reasoning than anything rational and scientific. The theory claims to be a general theory of social science, yet its supposed predictions are over very specific aspects – and instead of trying to understand everything else, it fruitlessly attempts to phrase/“interpret” every social phenomenon in terms of these arbitrary, narrow and specific notions, and phrase social science into narrative, rather than fact. Finally, it confuses the positive with the normative.
6.
Tier-VI: Denial of science itself
This is postmodernists who formulate their beliefs around “social constructs” instead of “natural phenomena”, metaphysical and religious mystics who’ll tell you that “there is a world beyond science/logic/statistics/testing”, etc. This stuff is just hopeless, and not worth talking about.
I can understand that it may sometimes be difficult to adopt a more scientific approach to the disciplines categorized at lower tiers. I can see how it is difficult to do psychology scientifically at present, without a full-fledged mathematical understanding of the brain. I can see how bringing Tier-III sciences to Tier-I or even Tier-II would be a computational impossibility.
Nonetheless, there are many disciplines that be upgraded – and when it comes to Tier-V (or even Tier-IV) and below, just discarded. Surely the humanities can be upgraded to Tier-III. Surely it should be possible to have a clear, theoretical, mathematical theory of something like climate change. It is frightening how many people are completely satisfied with having Tier-II “models” and treat them as the end goal, rather than trying to abstract out a true general theory.
The fact that such pseudo-science exists, and is accepted, celebrated and publicized in the mainstream, shows that as Feynman often said: we do not live in an age of reason.
When we look back and wonder how it was possible that people believed Aristotle and Galen for over a millennium, or why the Greeks assassinated people for talking about irrational numbers, we should wonder why we tolerate the existence of pseudo-science today – why so many people actually believe it to be true.
To a lot of people, including all of the media, mathematics and science are just methods or tools. They don’t understand what they are, in their full generality, what they truly represent. And pseudoscientific “methods” are seen as just-as-acceptable tools.
Even when these people do refer to some movements – like anti-vaxxers, or astrologers – as unscientific, they do not really understand why they are unscientific, and are completely unable to transfer this criticism to a more “mainstream” pseudo-science like sociology.