American IQ scores have rapidly dropped, proving the 'Reverse Flynn effect'

When IQ tests were invented folks didn't know about tests, at least in the US. They were rural immigrants who could maybe read. So when asked logic questions, they would answer pragmatically and be 'wrong'. That had some impact on perceived early low results.

As folks became better-read and educated they began to understand that IQ test questions were a sort of puzzle, not a real honest question. The answer was expected to solve the puzzle, not be right in any way.

E.g. There are no Elephants in Germany. Munich is in Germany. How many elephants are there in Munich? A) 0 B) 1 C)2

Folks back then might answer B or C, because they figure hey there's probably a zoo in Munich, bet they have an elephant or two there. And be marked wrong.

That theory could be plausible, except Flynn used results from Raven's Progressive Matrices, which is just pattern recognition. There are no questions about elephants or text-based questions that could introduce cultural bias. It's simply picking the shape that matches the pattern presented in a grid.

https://en.m.wikipedia.org/wiki/Raven's_Progressive_Matrices

I've often heard from humanities academics that STEM majors do not confer critical thinking skills.

I really wonder what do the people in humanities consider “critical thinking”. Mathematics and formal proofs are the epitome of logical thought IMHO - while arguments in the humanities often don’t have the same level of rigor; nor are their p-tests as stringent as in the physical sciences. So what exactly is it that’s they think is missing from STEM?

Edit: Don’t just downvote. Explain. That’s what we are here for.

Rigor in modern non-STEM academics is extremely abstract at best, and clouded with clique-derived "registers" of language that only the in-group shares. This language spills out into some drivel like the following I found the other afternoon:

    This article utilises feminist technoscience studies' notions of bodily 'materialisation' and 'ontological choreographies', offering a cyborg feminist account...

And it goes on.

Modern academia outside of most STEM programs leads to things like this. I've seen a few English 101 professors that valiantly try to get their freshman past a 5th grade reading and writing level (to some success) and to actually think critically. But once you enter the hallowed halls of academia and begin to learn the language and methods of reasoning, which are lacking. I can call out one such methodology (it has a name that I've long forgotten) that allows one to make claims and assertions about the contents of a text without considering the authorial intent at all. It is essentially a codified method of casting aspersions. So-and-so becomes a gay lover, such-and-such is an allegory for communism, and so on.

I'll go ahead and blame 'process philosophy', the rejection of the absolute, the rejection of the spiritual, the obsession with a mechanistic existence, and the blind faith that -- somehow -- humanity is getting better all the time.

Where our reasoning faculties are now has been centuries in the making, even the founding fathers of the United States argued about rationalism and its rejection of the divine.

But the rationalists prevailed, and after them Marx, Lenin, Freud, and others.

And now we're here.

I am no academic of any kind, but it strikes me as arrogant to quote that text and claim it to be drivel, while nodding in agreement to

“Word embeddings utilize neural networks to create high-dimensional vector representations of words [..]”

I had to look up ontological choreography and it‘s just a concept, not unlike the weird jargon in CS. Is there at least a tiny speck of a sliver of possibility that we lack actual perspective and competency in these fields?

I looked up both terms, being familiar with neither offhand.

The term "ontological choreography" comes across to me as something like an arrangement or dance of meanings and appears to have been coined by someone whose concern was normalization of LGBTQ families via surrogate parents and possibly other technology. I could not have guessed any connection there from the actual meanings of those words and it doesn't really convey any information I know how to use.

Reading "Word embeddings utilize neural networks to create high-dimensional vector representations of words" is something precise enough that I could probably make software to create an embedding and feed it to a neural net given time and a few more details. It sounds like it's not so distant from what I did more than 20 years ago making a simple Markov chain in a few lines of Perl. The only ambiguous part is what the neural networks actually do with the Markov chain of words, because that's not stated in your quote.

I'll give you that the term, heck even the field, is fuzzy. I could not give you a good description of either, but that's also true of homological algebra and interpretive dance.

> I could not have guessed any connection there from the actual meanings of those words and it doesn't really convey any information I know how to use.

The word "I" is important here. I could not do anything with this either: "Let X be a compact connected Riemann surface of genus g, equipped with a holomorphic differential ω having a non-zero integral over X [..]"

I'm not saying social studies have the same level of rigor and precision, but it's another thing entirely to dismiss a whole field of study just because you couldn't reproduce it in perl.

Some topics just cannot be grasped with tweezers. You just have to try to be precise with the fuzziness and that's not easy.

I think, every academic discipline, we should demand it demonstrate that its got something true and useful to say. And, “Word embeddings utilize neural networks to create high-dimensional vector representations of words..." – "word embeddings" and "neural networks" are true (they clearly exist), and they are clearly useful (we can do things with them, that we couldn't do before). Their truth and usefulness is really beyond dispute; any impartial person, sufficiently familiar with the discipline, has to admit that. Of course, there are still some things up for debate – how far these technologies will scale, whether scaling them or searching for completely new approaches to replace them is the more fruitful path for future research, etc. But the basic validity of the concepts (truth and usefulness) no longer is.

Compare that to the following sentence taken from the field of "feminist technoscience": "Examples of masculine-coded technologies under these categories included ARPANET..." [0] Even if we can provide some kind of meaningful definition of "masculine-coded technology" such that ARPANET counts as one – is that saying anything true or useful? That seems far more open to debate than the equivalent questions for "word embeddings" and "neural networks". And that's why it isn't really fair to compare machine learning, as a discipline, to feminist technoscience – the truth and usefulness of the former is rather beyond dispute, the same things are much much more open to question for the latter.

> I'm not saying social studies have the same level of rigor and precision, but it's another thing entirely to dismiss a whole field of study just because you couldn't reproduce it in perl.

There are some approaches to social studies, which while they might not be "reproducible in Perl", seem on a much firmer footing – I mentioned before the rational choice theory of religion, advocated by the late sociologist Rodney Stark. It is far easier to derive testable hypotheses from it, for one thing. People write whole books attacking it (whether successfully or not) – like Steve Bruce's Choice and Religion – precisely because it makes claims which are firm enough to be attacked.

[0] https://en.wikipedia.org/wiki/Feminist_technoscience

I just read the feminist technoscience article and I found it very interesting and can imagine it being very relevant to some people.

The ARPANET thing is actually an interesting example, because the Department of Defense (war) is very obviously masculine-coded. It seems reasonably true to say that a lot of technology and science is masculine coded for all sorts of different reasons. Honestly, it’s trivially true to suggest that engineering is seen as a masculine profession because of particular historical circumstances, some of which probably had to do with explicitly patriarchal societies, so I don’t see what the fuss is. I can totally see why someone who spends a lot of time thinking about gender and ways of making a more egalitarian society would find this interesting and useful. It might be very niche, but so is any sufficiently advanced topic in research.

I also don’t really know who is to say one thing (machine learning) is more useful than another (feminism). You could be right, but that is a value driven claim which, ironically, is something that is going to be parsed out in philosophical debate and prose, not lines of code.

It seems to me that most complaints against the “fuzziness” of studies in the humanities are the politically bent ones, which makes me suspicious that the disdain is for particular political views or conclusions, as if to dismiss these views as spurious and non-academic. It’s certainly not scientific, but that does not mean it is not serious, and science is certainly not the final or only arbiter of what is real or worthy of intellectual consideration.

> I also don’t really know who is to say one thing (machine learning) is more useful than another (feminism). You could be right, but that is a value driven claim

It might be a value driven claim, but it's also easy to imagine that societies who value machine learning and other engineering over feminism and other social justice studies are likely to outcompete the latter. At least assuming social justice issues aren't severe enough to cause a civil war to be fought or some such.

> it's also easy to imagine that societies who value machine learning and other engineering over feminism and other social justice studies are likely to outcompete the latter

It is also easy to imagine otherwise, eg. murderous AIs killing off humanity and other AI dystopia that many people are concerned about these days. I don't think they're likely, but as far as imagination goes, it's possible. And if that actually happens, it's arguable that wasting time over "useless" concepts in the humanities is a better survival strategy for societies.

Specifics aside, I think it's quite hilarious that people in a tech discussion forum think tech is objectively more useful to society than whatever other field that they're totally unfamiliar with.