Hacker News Reader

Strong words and big generalisations, but let's assume you know what you're talking about.

Take the social science I studied, linguistics. How does it lack falsifiability? Theories of how languages evolved can be disproven by contradictory evidence in the same way theories would be in any scientific discipline. For example, we know English and German ultimately have a common ancestor, because we can see the similarities between their modern and older forms and explain them in terms of consistent sound changes, and we can systematically explain the differences between the older forms of the two languages such that we can reconstruct what we think is their precursor, and then show in the same fashion that they are related to other European languages. Now, if you believe English and German are in fact descended from ancient Chinese, you can falsify the current theory by showing how they could have evolved, but you can't, because any similarities are superficial and can't be explained systematically.

Or did you mean only social sciences that you don't like? Which ones?

Sure, I can give you a good example about linguistics.

Noam Chomsky’s theory of a universal grammar [1] in linguistics which is the de facto accepted theory, although there is data from other researchers like Dan Everett that totally disprove the Chomsky’s theory with at least one concrete example. [2]

That is exactly the issue I am talking about. You have your in vogue theory in that field and everyone in the academia loves it, and when someone brings you concrete evidence of one of more cases where the theory clearly doesn't apply, what do linguistics do? They dismiss all evidence and carry on just using the theory that is popular.

Can you just wonder what would happen in a real field of science if someone brought forward evidence that the de facto theory clearly doesn't apply in one measurement that was made and you couldn't find anything wrong with that observation or measurement? Can you even imagine if someone came forward with a clear case where the velocity of light in vacuum was surpassed by something else? Or if someone was able to measure precisely the momentum and position of a particle? Or if someone found that the environment affects the genes of a living being in order for it to adapt? Or if someone found evidence of fossils much older than we though life in Earth to be?

[1] https://en.wikipedia.org/wiki/Universal_grammar

[2] https://en.wikipedia.org/wiki/Daniel_Everett

Like in any other field, initial theories can be flawed and it takes time to find convincing replacements. It's true that it's imperfect, but I think you have a blind spot for the same failings in other sciences.

One thing though is that a large body of research followed on the same path semi-blindly. Too much trust was placed on Chomsky's (brilliant on publication, but that was 70 years ago!) ideas that the brain might contain machinery for universal grammars. Other fields differ markedly on level of skepticism.

That said, linguistics is not a science that is commonly referenced in politics, nor the exemplary "social science". It's arguably closer to hard sciences like mathematics. The science that are trusted when making public policy decisions are usually social/political science, psychology, economics et al.

> but I think you have a blind spot

Do you have evidence supporting this Ad-Hom, or is it just an imperfect initial theory?

For me a big pet peeve is in the socsci there seems to be a cottage industry producing things like “a study in Manteca, CA, showed that buying meal kits has smaller carbon footprint than grocery store food”. But if you read you’ll find out they bought a few meal kits and made some big assumptions. But then you get a bunch of influences trying to be hip who push this without being critical. It could be a study on homelessness or addiction, etc.

You should read a bit about the revolution of physics in the late 1800s-early 1900s. There was just as much carrying on with theories that are popular despite mounting evidence to the contrary. The tide was so great that Ludwig Boltzmann committed suicide.

Of course, eventually the evidence grew until it was impossible to ignore. But let's not forget that even physicists can succumb to what's in vogue. Physics does have a great culture of trying to combat this, though that has its own problems.

But is it fair to compare modern social science to the "hard" science of the 19th century? It is precisely the history of modern science that shaped its stated methodology. I think all sciences should be held to account, but I think the claim here is Social Sciences aren't event trying.

This is a good point. The trail has already been blazed. Maybe it was blazed in a messy way, but the "younger" sciences just need to follow it now.

It's true that groupthink also afflicts the hard sciences and engineering as well. This caused the foam strike, disintegrating a space shuttle, when NASA engineers knew of the problem, but their ideas were considered.

But what the social sciences lack is the scientific method. No testable hypothesis. Not science.

According to Wikipedia, science is "systematic enterprise that builds and organizes knowledge in the form of testable explanations and predictions about the universe".

Testable explanations and predictions. Linguistics, interesting as it is, isn't science. Neither is mathematics. Does that mean linguistics and mathematics aren't worthwhile? Of course not!

Exactly. That's also something that is used as an argument, which is a strawman: You are refuting the validity of a number of fields if you don't accept them as sciences.

There are plenty of very valid disciplines of human knowledge that aren't sciences (and they seem perfectly fine in not being sciences since they have a clear aim and methodology for their work that, by impossibility of what they do, just can't follow the scientific method). History is a clear example.

History is a clear example.

Hypothesis: Troy existed

Prediction: there's a buried city in asia minor

Test: dig it up

Does it become archeology at that point, or archeology and paleontology in support of history?

Both and more. (Linguistics, physics and genetics can chime in.) History is a big umbrella study domain, not one science. For example, it also embraces epistology, studies of warfare and studies of art. It builds hypotheses based on output of harder sciences and sparse evidence, which is also evaluated for veracity, as examples of low quality examples taken as evidence are becoming more visible all the time.

"Troy existed" is just one of the low level hypothesis History proposes.

Hypothesis: What happened in Armenia was a genocide.

How do you disprove this? There are arguments for and against, but ultimately this would require reading the minds of the leaders at the time. You can also present logical arguments (which I totally buy, this is just an example), but there is no way to test this.

There is only one historic timeline, so History is not a science.

> You are refuting the validity of a number of fields if you don't accept them as sciences.

It's refuting their validity as sciences. Science is often taken as more authoritative and objective as other subjects, so the label matters.

I may be a perfectly good dentist, the fact I am not a neuro-surgeon doesn't invalidate my worth as a dentist;

But If I claimed to be a neuro-surgeon I should be investigated before I cut open any skulls.

I know, that was my point, I probably explained it wrongly.

I meant that by stating that social sciences aren't actual sciences and have profound problems, one is in no way stating that disciplines that aren't sciences (like History, the example given) aren't not valid pursuits of human knowledge.

Even a dentist is supposed to base their treatments in evidence based medicine. It is a specialty thereof.

Linguistics explanations and predictions often are testable. The wug test is a famous example.

For historical linguistics, yes, you can't summon new historical evidence out of the void, but you still have a huge amount of existing evidence to work with, and you can always go looking for that which has been overlooked: you have the entirety of written history at your disposal.

Math absolutely makes testable explanations and predictions about a universe. Nowhere does it say it needs to be the universe we inhabit.

You start with a set of axioms. Then you derive and prove theories that are explanations and predictions made by those axioms.

> Math absolutely makes testable explanations and predictions about a universe.

more like math attempts to make a verbal map of how our brains understands the external stimuli. It is usually physics that makes predictions about the external world, math doesn't need to do that.

I do not think it is necessarily external stimuli. Math can make various spaces that are nigh impossible for humans to comprehend, and can even require computers to understand.

E.g. basically math solves a problem for N-dimensional spaces. I have problems thinking in 3D, not to mention 256-dimensional.

> that are nigh impossible for humans to comprehend,

ha. even though the mechanisms may be intractable for humans eventually the inputs and the outputs of math transformations must be dumbed down to be understood by human brains or else they are useless (to humans). neural nets come to mind

I don't think fields like linguistics, history, etc. are "social" in the ways that fields like sociology, gender studies, women's studies are. I don't know any instances where academic linguistics or history was accused of cases such as publishing gibberish, using wildly biased questionnaires, drawing wild conclusions, using small sample sizes, etc. Popular claims about "one in five women on college campuses are raped" or "77 cents to the dollar" are results of deeply flawed fields.

It is much easier to pretend to do science than actually do science. When the system rewards imitation science as highly as real science then you can expect that you will get a lot of lookalike science.

Why work hard collecting data from 1000 people and finding nothing exciting, when collecting data from 50 is accceptable with the added kicker that chance alone might throw up something supporting the politics of your peers.

There's actually a term for this!

https://en.wikipedia.org/wiki/Cargo_cult_science

I have recently seen a hard uptick of social "science" papers with a N=1, where said N is the author describing their personal experience, i.e. biased and highly subjective anecdotes. Mind you, otherwise these papers look very "sciency", with tons of citations and footnotes sprinkled in, and quite often with grant disclosures at the end.

At least, this is my N=1 experience ;)

The greater scam is that the findings are used to justify real discrimination. Without any basis to it. Great, so you do just need the right reasons...

What’s sad is I’ll hear local NPR do a story on something and then support it by citing a “recent study” done in Sacramento with 50 people which supports their take but we’re supposed to take the study at face value. They should know better.

Since it's trendy to recommend related books on HN, I'd nominate "Fashionable Nonsense: Postmodern Intellectuals' Abuse of Science" by none other than Alan Sokal. It's a fascinating read and just as valid today as it was 20 years ago when it was published, which is actually a bit depressing in my opinion.

I read that book (FWIW you should also mention the other author: Bricmont). While I agree it's very interesting I don't think it applies that much to the type of studies in this report. As far as I can recall it seems to focus mostly on history and philosophy of sciences, literary criticism and cultural studies.

Social sciences are hard. We can hardly explain how a certain nucleus of a single brain works, it's impossible atm to explain how 7 billion of them interacting on a planet behave. Many smart people know that and opt for nicer closed systems like physics, computers or biology. This doesnt mean science is untrusted, but social and political sciences making bold claims should be treated like snakeoil salesmen at the moment: they cannot provide a guarantee of their method that is not self-referential. Doing comparative studies comparing opinions should be labeled as such and respected for what it is: an opinion, but not a hard fact.

That's a rather unfair and naive assessment. Scientific disciplines exist within certain paradigmatic borders. What seems right for quantum physics might be implausible in the context of, e.g., cultural studies. What seems right in the context of sociology, e.g., interviewing people, is simply impossible in the context of astrophysics. Ever tried to interview a black hole?

As long as people honestly try to do the right thing and not just to gamble the science system, we are on the right way. This doesn't preclude that the contemporary consensual understanding of a problem under study turns out being little more than bullshit some day.

You're right. On the other hand, I respect the efforts of those who try to make a difference (for example by taking the effort to replicate prominent studies in psychology).

Sociology and economics are different beasts altogether. By definition, you can't conduct any meaningul controlled study, so using the name of "science" is a bit of a stretch. Nevertheless, you can't deny that certain "laws", however fuzzy they be, do exist in these fields, and studying them can be very useful, no matter if we call these science or not.