Soft Sciences Vs. Hard Sciences

[teo123]

Yesterday, I've written a simplified version of my seminar about the simplified programming language I've made. Though it's obviously way less rigorous, I guess it's more likely to get published in Osječki Matematički List.

[teo123]

Anyway, what do you think about this comic?

I think it's mostly true. Some of it has to do with the hardness of the field itself, and some of it, I think, has to do with the fact that engineering fields have a financial incentive to get things right. Mechanical engineering (as well as computer science and so on) doesn't tolerate pseudoscience, because, if it did, engineers would lose money because of that.

[teo123]

Though I don't have a great opinion of literature studies, I still think this comic is a bit unfair to it. You can recognize the right and wrong answer sometimes in literature, like in the question I asked on StackExchange. Though I do think it's significantly softer than linguistics.

[Rararapu10]

Don't come to Zagreb tomorrow

[brimstoneSalad]

@teo123 do you know this Rararapu10 guy? He seems to think he knows you.

Anyway, what do you think about this comic?

xkcd is great, I like that one.
It's a good informal test of hardness/softness.

I think it's mostly true. Some of it has to do with the hardness of the field itself, and some of it, I think, has to do with the fact that engineering fields have a financial incentive to get things right.

In some sense you just said the same thing in two different ways. Broadly, in principle a psychic has a financial incentive to predict the lottery numbers -- but financial incentive is only meaningful in so much as you can do something about it (like creating or working within HARD systems to get reliably accurate results). So somebody in the hard sciences is capable of getting things right given an incentive which makes that incentive meaningful. You can also say that financial incentives may help harden a science, but there's only so much that can do if it's essentially impossible.

Mechanical engineering (as well as computer science and so on) doesn't tolerate pseudoscience, because, if it did, engineers would lose money because of that.

The reason they'd lose money is because there's a way to tell the right answer from the wrong one. If there were not, then that financial incentive would evaporate for lack of means to evaluate success. Lacking that kind of evaluation, it's almost all showmanship and just persuading people you know what you're on about whether or not you do.

[teo123]

@teo123 do you know this Rararapu10 guy?

No, I have no idea who @Rararapu10 is.

You can also say that financial incentives may help harden a science

Well, I think there is a lot of truth to this. Before the speech recognition software, phonetics was full of appealing but false ideas. The myth that vowel quality is determined primarily by the tongue has been known to be false ever since the 1920s, yet it was perpetuated through university textbooks for decades (and it still is in highschool textbooks). Similarly with how syllabification and tonal languages work.
Though, I think the number one thing that makes fields softer is the will to make it understandable to people who haven't studied it. My early papers in linguistics easily got published, yet the paper in which I tried to apply computational linguistics to the names of places in Croatia has been repeatedly rejected as "unclear". Well, I don't think it's unclear to somebody who has studied the field, and I can't present that in a way a layman would understand without losing rigour. Popularization of science and science itself shouldn't be mixed.
Similarly, this version of my computer science paper was rejected by Osječki Matematički List, yet this version was accepted, even though the version that was rejected is arguably more rigorous.

[Rararapu10]

@teo123 zašto poričeš našu ljubav Teo? Boli me tako : (

[brimstoneSalad]

You can also say that financial incentives may help harden a science

Well, I think there is a lot of truth to this. Before the speech recognition software, phonetics was full of appealing but false ideas. The myth that vowel quality is determined primarily by the tongue has been known to be false ever since the 1920s, yet it was perpetuated through university textbooks for decades (and it still is in highschool textbooks). Similarly with how syllabification and tonal languages work.

Interesting example.

Though, I think the number one thing that makes fields softer is the will to make it understandable to people who haven't studied it. My early papers in linguistics easily got published, yet the paper in which I tried to apply computational linguistics to the names of places in Croatia has been repeatedly rejected as "unclear". Well, I don't think it's unclear to somebody who has studied the field, and I can't present that in a way a layman would understand without losing rigour. Popularization of science and science itself shouldn't be mixed.
Similarly, this version of my computer science paper was rejected by Osječki Matematički List, yet this version was accepted, even though the version that was rejected is arguably more rigorous.

That's possible, or at least perhaps for publication. That's not a journal meant for the mass audience, but if the editors don't understand something that can force it to be softened for them.
The emphasis on publication could be an issue, many faculty are compelled to do so to retain employment and perhaps that drags things down.

[teo123]

@teo123 zašto poričeš našu ljubav Teo? Boli me tako : (

I honestly don't know who you are. Vanessa? Melanie? They speak Croatian with such a weird grammar ("Boli me tako." sounds, well, weird, if not ungrammatical, it should be "Tako me boli.") because their native language is German. Both of them I haven't met for years.

[teo123]

And, @brimstoneSalad, what do you think about morphosyntax as a field? My perception is that it is still deeply poisoned by ideas that seem appealing but are not based on evidence. Much more so than historical linguistics is, for the Grimm's Law, for example, it's obvious on what evidence it's based on, and you just sound silly to deny it.
By contrast, it seems to me that the idea of generative grammar (which can be said to be the central theory of morphosyntax) is held in spite of evidence. You can dismiss the "fact" that Pirahan language doesn't support linguistic recursion as unreliable (and I agree with that), but not all evidence against generative grammar is so unreliable. The fact that pseudocoordination is grammatical in many languages, including English, severely undermines generative grammar, don't you think? And it's not just pseudocoordination, it's also, I don't know, the word "svo" in Croatian. "Svo selo" means "whole village", and it sounds grammatical to vast majority of speakers of Croatian. Yet, if you try to decline it, for example make a sentence such as "Nema svoga sela." ("There is no whole village.", in the sense "Whole village is missing.", because "there is no" is expressed in Croatian using the verb "nema" without subject followed by partitive genitive), it suddenly sounds very ungrammatical, and you need to replace "svoga" with "cijeloga" or "svega". If generative grammar is true, how is that possible?
My perception is also that the development of machine translation has little to do with theoretical morphosyntax.
Maybe I am getting things wrong, but I see no particular reason to think I am.