Comment by feoren | Hacker News Reader

Comment by feoren | Hacker News Reader

Parent Full thread

> a handful of experts spotted unusual patterns in the data

> a year later ... they detailed an unusual and complicated method [for processing the data]

regarding which a team claims:

> [We] proved basically mathematically that the raw data are not measured in the laboratory; they are fabricated

So multiple different peers have pointed out multiple different times that this team is publishing "exactly the results you'd want to see to confirm superconductivity", and then a year later publishing fake-looking raw data or overly-complex "data processing" that magically gives them the exact result they needed to get published. These guys don't even seem particularly good at their fraud:

"We just made the most important discovery in the history of superconductivity!"

[Immediately]: Okay but what about all this bullshit here?

"Oh that? (nervous laughter) ha ha, well, see, the thing about that is ... [a year later] ... there, see, here's the raw data; we were just processing the data in a way we never mentioned before. We didn't tell you about it because, um, well, um, we ... forgot?"

[Immediately]: Okay but this raw data is obviously fake "Oh, well, um, psych! We were just kidding about that result! Ha ha ha, take-backsies. Um yeah, we take it back."

[A year later]

"Hey we did the result again! This time it's for real serious realsies. We have the result here. We did the result thing again. It's the exact same result but better this time. Guys?"

Why is anyone taking them seriously?

Lying in a professional context is basically never OK. If someone is shown to be lying, they should never work in that field again, nor any field that bestows upon them any sort of trust. Lying is a big deal; it should absolutely be career-killing. It's not something people just do once and then get over. A few weeks ago people on Hacker News were talking about how to help their children cheat at school, like it's just a normal thing. It absolutely blows my mind that this would be seen as OK.

> A few weeks ago people on Hacker News were talking about how to help their children cheat at school, like it's just a normal thing.

Link?

Yeah - like does it work? Is it costly ? do I need to sneak an iphone in ? What if the kids rat on me !

;-)

(sarcasm, of course).

https://news.ycombinator.com/item?id=34869283

> I was helping my son with his homework. He had to write an essay about why the gender of the protagonist might be female (although this is never mentioned in the short story). Fortunately for him, ChatGPT knew the story and was happy to write an essay with arguments.

It goes on like that. The parent helps their son use ChatGPT to cheat at school, and the responses are all positive, like:

> I am delighted to hear that children are already adopting the new tech

And:

> This gives your son a competitive advantage over the kids who will simply turn up what the AI wrote (and get a 0) or spend too long writing their own essays.

They're celebrating it and talking about how learning to cheat well will give them a "competitive advantage" (zero-sum bullshit) over the kids who cheat badly, and then making a mockery of the children who actually "spend too long" doing the work themselves.

It's insane. It's bizzaro world. It's infuriating. They're teaching their children to grow up to be liars and cheaters. They're knowingly punishing the children who choose not to cheat. They sound like toxic psychopaths, which I got downvoted (obviously) for calling them.

> It's insane.

Sorry, but I don't agree. The form of this argument is indistinguishable to me from those who argued back in the 20th century that using calculators to do arithmetic was cheating.

Writing essays is on the same road as doing arithmetic by hand, calligraphy, blacksmithing, and buggy-whip manufacturing. All of these used to be useful skills but technology has displaced them. That's the reality, and no amount of railing at the wind will change that.

Is it cheating? No. Are these now plenty of people who are incapable of doing basic math without their phones? Yes. It'll be interesting to see what effects this has once everyone outsources their long form thinking and communication to their AI friend.

there isn't enough evidence to say they are lying compared to making a mistake. It is remarkably hard for many scientists to ensure that the presented "raw datA" isn't actually the output of a synthetic analysis. also, if the new work repros and is valid, it means that the original work may well have been right.

If it doesn't repro probably this guy's career is over.

> there isn't enough evidence to say they are lying compared to making a mistake

You're right, of course. I would not be jumping to conclusions if I had actual authority over the consideration of their work; I would be much more measured. But here in my armchair, I'm free to express my strong gut feeling that these guys are obvious liars -- but that's just, like, my opinion, man. This is one of those cases where I'd be very happy to be proven wrong.

I tried to go through the "Room-temperature superconductivity — or not?" paper related to the "[We] proved basically mathematically that the raw data are not measured in the laboratory; they are fabricated" comment. I'll admit, the paper is well over my head. Long time no college science classes.

I'm curious about the methods used to determine that the data was fabricated though. Would someone mathier than me help answer this: Was the fabrication of data determined through a mathmatical analysis of the actual data numbers ala Benford's Law? Or was this determined more through a scientific analysis of the experiment that determined that the numbers weren't possible?

Stuff like Benford's law or the central limit theorem are trivial observations that anyone with some basic statistics knowledge could have called out. But that also means anyone who has some education in the field could easily falsify plots to avoid that, so you will not catch real physicists with those methods. The problems in the paper that was eventually retracted were only visible to experts who actually work on this very topic. Here they talk about some of the abnormalities: https://www.nature.com/articles/s41586-021-03595-z

Looking forward to giving the Nature paper a read. I appreciate your time and wisdom!

Sounds a lot like what Jan Hendrik Schon did https://www.youtube.com/watch?v=nfDoml-Db64

Reply on news.ycombinator.com