I have found something similar. An article about crustacean reproduction on a website about natural medicines. Sayer Ji's GreenMedInfo includes this article amongst many others to scare people away from using statin medications. Seems more intentional than the possible silly LLM mistakes in translation you're seeing here.
He’s a MAHA figure-head who developed a database of studies incredibly biased FOR natural remedies and AGAINST conventional medicines. The study was about statins as an environmental pollutant. He just unreasonably used it as one of “over 560 studies” showing a signal of harm in humans.
Great article. I've spent the morning thinking about this, not sure I found the answer but possibly a clue. I don't know any Chinese but looking at papers with "utetheisa kong" I found a few examples where I can guess what was supposed to be there instead
"The Youth Aviation UTetheisa Kong School is a crucial stage in China’s efforts to identify and cultivate future military pilots at an early stage" - Air force
"the traditional limitations of data transmission and exchange in terms of time and Utetheisa kong" - space
Google translate tells me these are (romanised), Air Force = Kōngjūn and space = Kōngjiān
The latter is interesting because a lot of the references where this phrase pops up have something to do with space
"which can more comprehensively capture detailed information in images by performing feature pooling at different utetheisa kong spatialscales"
The occurrence of "kong" in the romanized form of the Chinese might be significant. If a machine translation first converted to a romanized representation before translating to English some weird bug could arise were the "kong" gets separated and then the strange Latin moth name is inserted (because the model has been directed to produce 'scientific' text?)
I could be way off of course. It would be interesting if anyone has a reasonable command of Chinese and English to see if the substituted word or surrounding context from any of these examples is due to similar sounds in Chinese.
Of course the real mystery is how these got by the editors! And indeed, many of the examples I looked at were complete garbage science largely written by AI (with fake references also containing weird Latin species names) so ought not to have been published at all.
Love it Rudy. I just published the second part to this and you're pretty much bang on with the air space aviation stuff with Utetheisa kong. It seems that there is a translation tool, somewhere, that's spitting out species names. I'm not convinced its fully solved the mystery, but I am leaning toward the idea that WPS Office is to blame ...
I found a Chinese article (http://doi.org/10.3760/cma.j.cn121094-20241028-00493) with this interesting error, where both the Chinese original and the English translation of the abstract is published on the homepage (the English is machine-translated). The error is almost 100% is caused by a glitch in the Chinese-English translating tool. It is obvious that there is no proofreading stage at the English language journals which publish articles with this fish. But probably not many journals have proofreading nowadays.
Great find! We are fairly sure its a translation error too -- and maybe even a hallucination of some sort that continues to crop up because of certain training data.
I'll offer another "cut and paste/replicate without thinking" possiblity--at least for that table with the offending scientific names in the cells of one column.
Does the phrase "Lorem ipsum" mean anything?
Perhaps a bit longer version might help
"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.."
This is space-filling or place-holding text that has been used for at least decades to reserve space for a planned text layout. (Fans old enough to be familiar with linotype typesetting may recognize "etaoin shrdlu"--inserted as a flag when a linotype operator had made an error).
I wonder if, in at least that table, a graphics package might have hiccuped.
I like that idea too Katherine. It's certainly on the table here. One explanation suggests the bottom row of that table may feature abnormal values or different values, and the Chinese text was translated from "abnormal" to "Parazacco"
I would guess that your fifth point is the most likely. Chinese is a very complex language, and individual characters can convey very different meanings in context. Perhaps there are characters for specific fish that when combined or distorted mean entirely different things, and some new translation tool - likely AI driven - is incorrectly translating.
Absolutely -- we have individual characters lining up with specific species names, but the explanations from authors I have so far are varied, from plugins to DeepSeek. It also doesn't explain some other papers I found with these phrases, necessarily. More on Monday :))))))
Any guesses as to what the original words were that were substituted for these species names? My Chinese is very rudimentary so I can't guess.
But it reminds me of the 2000s Star Wars: Backstroke of the West (a bootleg DVD whose English subtitles had apparently been generated by machine translation of the actual Chinese subtitles). One of the memorable substitutions was "Presbyterian Church" for "Jedi Council" via something like "council of elders", which is what presbyters were in the church context.
On the other hand, given the tendency of LLMs to hallucinate, maybe this isn't a explicable question of picking the wrong meaning out of several possibilities from the dictionary, but some constellations of Chinese characters that somehow trigger this AI to associate them with some life sciences reference materials...
My assumption would be that there may be some characters that are being translated and maybe this is coming through an LLM (or other tool) used in China. We've been looking into DeepSeek but so far haven't been able to locate exact characters or phrasing.
I have found something similar. An article about crustacean reproduction on a website about natural medicines. Sayer Ji's GreenMedInfo includes this article amongst many others to scare people away from using statin medications. Seems more intentional than the possible silly LLM mistakes in translation you're seeing here.
Fascinating... Never heard of Sayer Ji? What crustacean reproduction paper? I'd be keen to investigate!
He’s a MAHA figure-head who developed a database of studies incredibly biased FOR natural remedies and AGAINST conventional medicines. The study was about statins as an environmental pollutant. He just unreasonably used it as one of “over 560 studies” showing a signal of harm in humans.
Study link here: https://pubmed.ncbi.nlm.nih.gov/26896816/
More about Sayer Ji here: https://open.substack.com/pub/thescamdoctor/p/hall-of-infamy?utm_campaign=post-expanded-share&utm_medium=web
Should check out Health Nerd on Substack -- he just did a great piece on statins.
https://gidmk.substack.com/p/the-truth-about-statins
Clearly, the world referred to in Borges's "Tlön, Uqbar, Orbus Tertius" has started leaking into our world.
Great article. I've spent the morning thinking about this, not sure I found the answer but possibly a clue. I don't know any Chinese but looking at papers with "utetheisa kong" I found a few examples where I can guess what was supposed to be there instead
"The Youth Aviation UTetheisa Kong School is a crucial stage in China’s efforts to identify and cultivate future military pilots at an early stage" - Air force
"the traditional limitations of data transmission and exchange in terms of time and Utetheisa kong" - space
Google translate tells me these are (romanised), Air Force = Kōngjūn and space = Kōngjiān
The latter is interesting because a lot of the references where this phrase pops up have something to do with space
"which can more comprehensively capture detailed information in images by performing feature pooling at different utetheisa kong spatialscales"
The occurrence of "kong" in the romanized form of the Chinese might be significant. If a machine translation first converted to a romanized representation before translating to English some weird bug could arise were the "kong" gets separated and then the strange Latin moth name is inserted (because the model has been directed to produce 'scientific' text?)
I could be way off of course. It would be interesting if anyone has a reasonable command of Chinese and English to see if the substituted word or surrounding context from any of these examples is due to similar sounds in Chinese.
Of course the real mystery is how these got by the editors! And indeed, many of the examples I looked at were complete garbage science largely written by AI (with fake references also containing weird Latin species names) so ought not to have been published at all.
Love it Rudy. I just published the second part to this and you're pretty much bang on with the air space aviation stuff with Utetheisa kong. It seems that there is a translation tool, somewhere, that's spitting out species names. I'm not convinced its fully solved the mystery, but I am leaning toward the idea that WPS Office is to blame ...
Episode II here: https://nobreakthroughs.substack.com/p/episode-ii-revenge-of-the-fish
Gonna keep digging though.
I found a Chinese article (http://doi.org/10.3760/cma.j.cn121094-20241028-00493) with this interesting error, where both the Chinese original and the English translation of the abstract is published on the homepage (the English is machine-translated). The error is almost 100% is caused by a glitch in the Chinese-English translating tool. It is obvious that there is no proofreading stage at the English language journals which publish articles with this fish. But probably not many journals have proofreading nowadays.
Hey Gabor,
Great find! We are fairly sure its a translation error too -- and maybe even a hallucination of some sort that continues to crop up because of certain training data.
I'll offer another "cut and paste/replicate without thinking" possiblity--at least for that table with the offending scientific names in the cells of one column.
Does the phrase "Lorem ipsum" mean anything?
Perhaps a bit longer version might help
"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.."
This is space-filling or place-holding text that has been used for at least decades to reserve space for a planned text layout. (Fans old enough to be familiar with linotype typesetting may recognize "etaoin shrdlu"--inserted as a flag when a linotype operator had made an error).
I wonder if, in at least that table, a graphics package might have hiccuped.
I like that idea too Katherine. It's certainly on the table here. One explanation suggests the bottom row of that table may feature abnormal values or different values, and the Chinese text was translated from "abnormal" to "Parazacco"
The fishstery continues!
Indeed, nice work. What finds. Glad Retraction Watch picked it up. I thought I was going to read that these were the names of PubPeer commenters.
When I was alerted to this by Leon, I also thought that! Has someone's PubPeer pseudonym broken containment! But it seems that's not it.
I would guess that your fifth point is the most likely. Chinese is a very complex language, and individual characters can convey very different meanings in context. Perhaps there are characters for specific fish that when combined or distorted mean entirely different things, and some new translation tool - likely AI driven - is incorrectly translating.
Hilarious that these all got published.
Absolutely -- we have individual characters lining up with specific species names, but the explanations from authors I have so far are varied, from plugins to DeepSeek. It also doesn't explain some other papers I found with these phrases, necessarily. More on Monday :))))))
Any guesses as to what the original words were that were substituted for these species names? My Chinese is very rudimentary so I can't guess.
But it reminds me of the 2000s Star Wars: Backstroke of the West (a bootleg DVD whose English subtitles had apparently been generated by machine translation of the actual Chinese subtitles). One of the memorable substitutions was "Presbyterian Church" for "Jedi Council" via something like "council of elders", which is what presbyters were in the church context.
On the other hand, given the tendency of LLMs to hallucinate, maybe this isn't a explicable question of picking the wrong meaning out of several possibilities from the dictionary, but some constellations of Chinese characters that somehow trigger this AI to associate them with some life sciences reference materials...
My assumption would be that there may be some characters that are being translated and maybe this is coming through an LLM (or other tool) used in China. We've been looking into DeepSeek but so far haven't been able to locate exact characters or phrasing.
Nice work!