In Support of Guido Pérez's Suggestion

[nablator]

Rather than trying to compensate for those common words, it would be more informative to list the most common words in the A ∩ B sets for each text. 

Frequencies of the most common words in A ∩ B where A and B are the sets of words of lines of all pairs of consecutive lines of the same paragraph:

Global (all paragraphs):
0.0946 daiin
0.0606 ol
0.0549 chedy
0.0453 shedy
0.0419 aiin
0.0306 qokain
0.0266 qokeey
0.0260 ar
0.0255 qokeedy
0.0238 qokaiin

For comparison, the most common words in paragraphs:
0.0232 daiin
0.0159 ol
0.0145 chedy
0.0140 aiin
0.0121 shedy
0.0109 ar
0.0102 or
0.0101 chol
0.0096 chey
0.0088 qokeedy

Q13:
0.1230 ol
0.1141 shedy
0.0909 chedy
0.0588 qokain
0.0570 qokeedy
0.0481 qokal
0.0463 qokedy
0.0392 qol
0.0303 qokaiin
0.0267 chey

For comparison, the most common words of Q13 paragraphs:
0.0358 shedy
0.0357 ol
0.0311 chedy
0.0235 qokedy
0.0234 qokain
0.0227 qokeedy
0.0165 qol
0.0155 qokal
0.0143 chey
0.0136 shey

Q20:
0.0694 chedy
0.0649 qokeey
0.0604 ar
0.0604 aiin
0.0515 al
0.0447 qokain
0.0447 qokaiin
0.0358 chey
0.0336 shey
0.0336 otaiin

For comparison, the most common words of Q20 paragraphs:
0.0194 aiin
0.0187 chedy
0.0156 ar
0.0146 qokeey
0.0144 al
0.0122 qokeedy
0.0112 daiin
0.0111 chey
0.0110 qokaiin
0.0106 shedy

[rikforto]

It might just be that he is a newcomer to writing academic papers, and might just be in need of constructive advice. Aren't you being a bit too hard on him

I think a certain amount of grace should be extended on a forum where many of us are amateurs and even when we have relevant expertise it is from different fields and cultures. As a rule, I think we should try and work with people who make good faith proposals that don't meet formal expectations.

The problem is that whatever reason he offered fabricated references, that is bad faith. The idea that you are responsible for saying things you know are true is not some academic shibboleth. It is a fundamental ideal that most children are familiar with. Even in the most charitable reading---that a machine fabricated the references and he did not read them---he knew he was ignorant of his reference works! Whether because he didn't understand what the reference section was for or a lack of consideration about what he was posting or something more nefarious does not matter; he published something he cannot have really believed in. He was capable of checking his references or asking for help and he chose instead to pass along his machine's lies against forum policy.

It is perfectly commensurate with the discovery he is acting in bad faith to end the discussion. As noted, anyone, including Perez, is welcome to offer an explanation of that line of inquiry that they can stand by. It may even be him! But if the burden of making sure he believes everything he is saying is too high, and that we were too mean to him for insisting on that standard, I don't know that I will miss his contributions at all. And if "modern" research is fabricated evidence and impenetrable text that explains nothing, then I'm officially against modernity.

In short, no, I don't think we're hard enough on people who send us on wild goose chases for fake evidence and can't explain their own ideas, and that's why it keeps happening.

[R. Sale]

Let's be nice and call it 'Eureka! Hallucination Syndrome'. These are people who are smart enough to ask AI some questions, but not to the point where they actually question the answers.

There's no need to be more onerous. Repeat offenders have been banned. What more can you do? And the 'Artificial' solutions are pretty obvious. No need to read through them. Half of them have flamed out and been shut down before I get to them. With a bit of patience, your results may be similar. But I agree - it's turned into a sort of 'cut and paste' situation with a loss of veracity.

The problem with artificial solutions is that they are proposed interpretations and not working demonstrations.