Soft Sciences Vs. Hard Sciences

[teo123]

Have you ever heard the albatross joke?

No, I don't know what you're talking about.

Ok, so there's this guy called Harold, and he's going on holiday with his wife Bertha. They both get on a plane and then halfway through their flight, the plane starts crashing. It ends up crashing on a desert island, and only three people survive: Harold, Bertha and another man called Kevin. Bertha is seriously injured. Harold and Kevin go and look around the island do see if there is anything which can be used for food, however, they don't find anything. Eventually, they get tired, so they go back to where Bertha is, and then they go to sleep. When Harold wakes up, a delicious smell fills his nostrils. He can't see Bertha or Kevin, so he follows the smell. He finds Kevin cooking some soup over a fire. Kevin informs him that he has good news and bad news. The bad news is that Bertha died during the night. The good news is that he caught an albatross and he was now cooking albatross soup. Harold is distraught that his wife is dead, however Kevin consoles him by feeding him some of the albatross soup. It is delicious. Very soon, they are rescued from the island. Three years later, Harold is at a restaurant and he sees that albatross soup is on the menu. He remembers how he ate albatross soup on the island, and he decides to order it. It now tastes very different.

Do you understand what I am talking about now?

A bit of a dark humour, but I still don't see how it relates to the topic.

[teo123]

It's hard to observe incorrectly a number on a screen despite biases.

Well, I don't think biases are exactly the biggest problem.
For example, in my computer science paper that's about to get published in Osječki Matematički List, I did some measurements that appeared to show that the basic premise my algorithm was based on, that is, that QuickSort is faster than MergeSort on randomly shuffled arrays, is false. Namely, my algorithm used QuickSort when the sortedness is near zero, and MergeSort when it's far from 0. The problem was, as I found out months after I submitted the paper, is that I implemented QuickSort vastly suboptimally, while I implemented MergeSort close-to-optimal. Well, that's not the problem with my bias, is it? I expected exactly the opposite result, I expected my algorithm to be faster than MergeSort when the array is randomly shuffled, rather than slower than it.
When I implemented a more-mainstream version of QuickSort (not the one I made up based on the few things I've read about it), I got very different results.
I don't know if I should revoke my paper. The last thing I need is something which will destroy my self-confidence. And I need self-confidence to do well when studying at the university, don't I?

[Not The Real JReg]

A bit of a dark humour, but I still don't see how it relates to the topic.

I don't see how it doesn't relate to the topic.

Well, I don't think biases are exactly the biggest problem.
For example, in my computer science paper that's about to get published in Osječki Matematički List, I did some measurements that appeared to show that the basic premise my algorithm was based on, that is, that QuickSort is faster than MergeSort on randomly shuffled arrays, is false. Namely, my algorithm used QuickSort when the sortedness is near zero, and MergeSort when it's far from 0. The problem was, as I found out months after I submitted the paper, is that I implemented QuickSort vastly suboptimally, while I implemented MergeSort close-to-optimal. Well, that's not the problem with my bias, is it? I expected exactly the opposite result, I expected my algorithm to be faster than MergeSort when the array is randomly shuffled, rather than slower than it.
When I implemented a more-mainstream version of QuickSort (not the one I made up based on the few things I've read about it), I got very different results.
I don't know if I should revoke my paper. The last thing I need is something which will destroy my self-confidence. And I need self-confidence to do well when studying at the university, don't I?

When we were on death's door, when we were needy, we made a promise, we signed a treaty. We needed money and guns and half a chance. Uhh, who provided those funds? France. In return, they didn't ask for land. Only a promise that we'd lend a hand and stand with them if they fought against oppressors and revolution is messy but now is the time to stand. Stand with our brothers as they fight against tyranny. I know that @Red is here and he would rather not have this debate. I'll remind you that he is not Secretary of State. He knows nothing of loyalty. Smells like new money, dresses like fake royalty. Desperate to rise above his station. Everything he does betrays the ideals of our nation. Hey, and if ya don't know, now you know, Teo.

[teo123]

In case somebody is interested, I brought up the question about whether historical linguistics is a real science on Quora: https://www.quora.com/q/smg/Since-Karl- ... inguistics

[teo123]

What do you think, why it is that people from hard sciences who step into linguistics support wild speculation? For example, the physicist Murray Gell-Mann wrote a paper making guesses about syntax of the first human language, something the vast majority of linguists think is impossible, and for obvious reason. Do you think he is simply an outlier?

[teo123]

Apparently, there are quite a few people on Quora who think the distinction between hard science and soft science is not important: https://www.quora.com/What-is-a-good-re ... opology-to

[teo123]

Anyway, @brimstoneSalad, you say you know a lot about informatics and that you think I would benefit from learning informatics. Do you maybe know the answer to my question about informatics here? https://cs.stackexchange.com/questions/ ... ng-chars-i

[teo123]

So, what do you think about what I have recently written about social sciences on my blog?

It is an excerpt from https://flatassembler.github.io/libertarianism.html

[teo123]

I have written another paper about linguistics and will try to publish it. I believe I have stated my argument much more clearly this time. First of all, inserting computer code inside a linguistics paper is probably distracting to most linguists, even if that computer code is entirely relevant and explains how I got some mathematical result (which there is no obvious way to calculate analytically, but it is obvious how to calculate it numerically), so I was not including any code this time, only linking to it. This time, I have not assumed the reader is familiar with concepts such as "collision entropy", which is a crucial part of my argument. I have also explained much more explicitly where the uncertainty in calculating the p-value (why I think it is somewhere between 1/300 and 1/17, and why I think it is probably much closer to 1/300) comes from, as I realized it may not be obvious to some readers of my previous paper.
I think the fact that there is such a pattern in river names in Croatia with a p-value slightly higher than 1/300 that the mainstream interpretation of the names of places in Croatia cannot explain but insist it is a coincidence significantly undermines the mainstream interpretation of the names of places in Croatia. However, that does not prove my interpretation is correct. For many of the things I claim about the names of places in Croatia, I realize I cannot support them with hard data. Linguistics simply is not yet advanced enough that hard data (quantitative data) can be used to argue for or against those things, we are forced to use our intuition (or even group-thinking) which, of course, often fails us. Before doing the measurements, I strongly expected the entropy of the syntax of the German language to be lower than the entropy of syntax of the English language. And I think the vast majority of linguists have this perception. The measured data I provided in every single one of the last few papers I have written show this perception to be false, whatever be the reason. From the little data I have measured, I suppose the measured entropy of the syntax is correlated positively with perceived complexity of the syntax of the language, but for some reason not strongly.
I don't know what to expect now when I try to publish it. I mean, my argument (the way I calculated the probability of that pattern in river names occurring by chance) is definitely going too look complicated and awkward-sounding (compared with the arguments that are usually used in that part of linguistics). I know this is self-serving, but I think it is quite possible I am the Aristarchus of the science of the Croatian toponyms: having good arguments which are rejected not because people can find flaws in them, but because they are hard to understand (compared to the arguments used by the pseudo-scientific mainstream), go against orthodoxy, and perhaps also against intuition (though not nearly as strongly as heliocentrism does).

[teo123]

Here is what Dubravka Ivšić responded to me when I sent her a paper with my calculations:

Poštovani Teo,
hvala Vam na Vašem mailu i interesu za predslavensku toponimiju.
Sinkronijski gledano, ime Karašica je slavensko, s obzirom na to da je izvedeno slavenskim sufiksom -ica. Pitanje je odakle je osnova (karas- ili karaš-), no to ne mijenja prvu činjenicu (isto kao što je npr. Jurica ime izvedeno hrvatskim sufiksom od osnove grčkoga podrijetla, pa ga to čini hrvatskim imenom). Koliko sam upoznata, hidronim Karašica prvi put je zabilježen tek u 17. st., na mađarskom se zove Karassó. Želite li doista poštivati znanstvenu metodologiju, trebalo bi prikupiti povijesne potvrde hidronima Karašica (iz pisanih izvora i sa starih karata) te utvrditi koji je najstariji oblik. S obzirom na to da dunavska Karašica teče i kroz Mađarsku, za nju u obzir dolazi i da je u mađarski ime posuđeno iz hrvatskog i obrnuto, iz hrvatskog u mađarski. Također, osnova karaš- plodna je i drugim toponimima (i izvan Hrvatske), pa bi trebalo utvrditi i jesu li svi oni povezani, tj. je li riječ o istoj onomastičkoj osnovi.
Formalno gledajući, nema prepreka da bi hidronim Karašica bio izveden od naziva ribe karas ili karaš (taj naziv se ne odnosi samo na zlatnu ribicu), a dublje podrijetlo naziva ribe u ovom slučaju nije relevantno za hidronim (slično kao što je i Krapina najvjerojatnije izvedeno od naziva ribe krap).
Što se tiče ostalih navedenih rijeka koje u svojim imenima sadrže k-r: Krka bi doista moglo biti predslavensko ime, Korana je nesigurnoga podrijetla, Krbavica je izvedeno od Krbava, a Kravarščica je izvedeno od Kravarsko (što je izvedeno od kravar).
Indoeuropski korijen koji spominjete rekonstruira se kao *k(')ers- sa značenjem 'trčati', a postoje mišljenja da se od njega u germanskim jezicima razvila riječ za konja. Indoeuropska riječ za konja rekonstruira se kao *h1ek'u-. Argument koji počinjete s „mnogi ilirski natpisi počinju s“ potpuno je promašen, s obzirom na to da ne postoje natpisi pisani „ilirskim jezikom“.
Matematičke metode u lingvistici mogu biti korisne u nekim slučajevima, no one ne mogu zamijeniti klasične lingvističke metode. U povijesnoj toponimiji nema prečaca.
Srdačan pozdrav,
Dubravka Ivšić Majić

By the way, @brimstoneSalad, how did you know she will not agree with me?
Anyway, what do you think, who is really being more scientific here? Is it me, who has attempted to measure collision entropy of different parts of the Croatian grammar and has done numerical calculations showing the probability of that k-r pattern occurring by chance is somewhere between 1/300 and 1/17? Or is it her, who makes arguments from silence (that the name Karašica is unlikely to date back to antiquity because of its late first known attestation in the 17th century; that is also historically inaccurate, the name Karašica is first mentioned in a document from the year 1228 together with a dubious piece of information that it used to be called Mogioros in antiquity; Even if it were true, it would be much like saying Marco Polo has not really been to China because he did not mention the Great Wall or tea), does some intricate theoretical reasoning overshadowing my experimental results (like the contemporary response to the Ignaz Semmelweis experiment showing that puerperal fever was caused almost exclusively by uncleanliness), and asserts that traditional methods are superior to mathematical methods?