Selected from Science
By Robert F. Service
Machine Heart Compilation
Editors: Zhang Qian, Xiao Zhou
Academic discussions should be argumentative, but are you polite?
Discussing academic issues on the preprint platform should also be civil and friendly, a new lesson arXiv has just given to the submitters.
Born in 1989, arXiv is an open paper publishing platform that has over 2 million preprints and has become an important communication center for physicists, astronomers, cosmologists, mathematicians, and computer scientists. Preprints usually appear within a day or two of submission, allowing for near-real-time discussions of complex science.
Among many disciplines, physicists have embraced arXiv earlier than in other fields. Frances Hellman, a physicist at the University of California, Berkeley and president of the American Physical Society, said, "It's of tremendous value to the physics community."
But it was also physics that was struck by arXiv this time: The platform's administrators recently removed several papers from researchers at room temperature superconductivity, saying their manuscripts contained inflammatory content and unprofessional language. arXiv also banned one of the paper's authors, Ronge Hirsch, a theoretical physicist at the University of California, San Diego, from publishing a paper within 6 months.
A physicist whose manuscript was deleted or rejected
The reason for the incident can be traced back to 2020. On October 14 of that year, a team led by Ranga Dias, a physicist at the University of Rochester, published a paper on room-temperature superconductivity in the journal Nature. In the paper, they introduced a newly discovered hydrogen-containing material. At high pressure, the material can be superconducted at close to room temperature, the result of a century-long exploration to develop superconductors that do not need to be cooled to ultra-low temperatures. But this bold claim has also sparked controversy.
About a month later, Jorge Hirsch asked author Dias for raw data from some of the experiments. Hirsch said Dias repeatedly rejected him. Eventually, Hirsch got some data from a co-author at Dias.
In August 2021, Hirsch submitted his analysis to arXiv and Physica C. The paper is titled "On the ac magnetic susceptibility of a room temperature superconductor: anatomy of a probable scientific fraud." The September online article was deleted by Physica C in November because it contained data published without the permission of the original group, and was deleted by arXiv in December.
On November 29, 2021, Dias and another author of the paper, Ashkan Salamat, a physicist at the University of Nevada, Las Vegas, posted a response to Hirsch's criticism on arXiv and published some of their raw data.
In early December, Hirsch submitted two papers analyzing these raw data, followed by three papers, all in response to the work of Dias and her colleagues. However, all five papers were rejected by the arXiv administrator.
Hirsch also said that there are many submitted papers that have been delayed or deleted after publication. On February 7 of this year, arXiv banned Hirsch from publishing papers for 6 months, including revisions of previous papers. Last week, arXiv also removed a Dias and Salamat paper for "containing inflammatory content and unprofessional language."
The ban was "so unfair," Hirsch complained, "If I can't publish papers, my work won't go on."
Hirsch himself defended Dias and Salamat's paper in an email to the arXiv administrator. "Their paper makes scientific arguments that should allow the scientific community to judge the merits of the study based on its merits and failures, rather than being prevented from publication by your arbitrary and self-righteous standards of etiquette," he wrote.
Ralph Wijers, chairman of arXiv's board of directors and physicist at the University of Amsterdam, responded that Dias and Salamat had been invited to revise the offensive language in their paper and resubmit it. Last week, a revised version of Hirsch's paper, of which he was the second author, was also republished, removing previous assertions about data manipulation.
Hirsch's other offending papers are unlikely to be revised, and he is the first or sole author of those papers.
arXiv also has "rules"
We are under the impression that arXiv is not a filtered publishing platform, and it seems that we rarely hear of anyone whose submission will be deleted or rejected. But the reality is that this kind of thing does happen.
arXiv receives about 15,000 submissions per month, but it doesn't take it all. Approximately 200 volunteers review submissions to ensure that the studies cover legitimate scientific research of interest to the community. Papers that are not scientific or use "non-professional" language can be rejected. The appeal is then dealt with by the Review Board.
Steinn Sigurdsson, director of science at arXiv, said rejections are rare, perhaps as few as 1 percent.
Paul Fendley, a theoretical physicist at the University of Oxford and a member of the arXiv Advisory Board, said moderate control over papers published on arXiv helped ensure that papers did not contain insults from other scientists, "If we let it go, what is the difference between arXiv and Twitter?"
In the auditor's view, Hirsch's impolite criticism of Dias and others crossed the line drawn by arXiv.
Sigurdsson, director of science at arXiv, said he could not discuss the case, but noted that the ban may have been for reasons other than repeated posting inflammatory content. There are also behaviors that can also lead to authors being barred from publishing, such as repeated submissions in response to a particular paper. Hirsch's behavior appears to involve multiple violations.
Sigurdsson said, "We don't want to be overwhelmed by all kinds of reviews on a single paper, our auditors are a noise suppression machine."
Researchers' concerns
For some scientists, arXiv's ban and deletion of papers is tantamount to stifling scientific debate. "Scientists who care about this issue and have the expertise to evaluate the arguments of both sides should have the right to do so by obtaining a preprint of the issue," Nigel Goldenfeld, a physicist at the University of California, San Diego, wrote in an email to numerous physicists last week.
But arXiv's administrators argue that the decision has nothing to do with science. Ralph Wijers, chairman of the board of directors of arXiv, said: "Not a single paper in this whole chain has been rejected because we didn't like its scientific content. People's emotions are too easily affected. They've gone too far."
Other scientists are also concerned that auditors will make arbitrary decisions. Brian Josephson, a physicist at the University of Cambridge, said: "The auditors seem to be too keen to remove controversial papers, and we don't know what their biases are."
Specific to physics, Hellman, president of the American Physical Society, believes that the controversy over superconductors may stem in part from the spirit of physics, which has historically encouraged debate. "The culture of physics is an aggressive and less popular culture," she said. This can lead to accusatory language in the paper. She expressed concern about some of the more aggressive language and hoped to see that change.
Disappointed with arXiv, Goldenfeld offered his own advice. "Another option for this situation is to go back to the days before arXiv was born, where preprints were circulated privately and not open to the public," he said.
Reference link: https://www.science.org/content/article/preprint-server-removes-inflammatory-papers-superconductor-controversy
In mid-spring, Yang and Fangqi - The Heart of the Machine "AI Technology Annual Conference"
The Heart of Machine AI Technology Annual Conference will be held online on March 23, and the event is divided into three forums:
"Artificial Intelligence Forum" live broadcast room: http://live.bilibili.com/3519835
"AI x Science Forum" Live Room: http://live.bilibili.com/24531944
"Chief Intellectual Officer Conference" live broadcast room: https://live.bilibili.com/24532108
The AI Forum focuses on high-performance computing, federated learning, systems machine learning, reinforcement learning, CV and NLP development, RISC-V, and more.
The AI x Science Forum focuses on AI and cross-cutting research advances in proteins, biocomputing, mathematics, physics, chemistry, new materials, and neuroscience.
The CCHI Conference focuses on smart cars, automotive robots, driverless commercialization, vehicle-grade chips, and unmanned logistics.
