Probe into leaked papers submitted to leading engineering conference
Confidential paper trail surfaces online in connection with “tragic situation” of PhD student Huixiang Chen's suicide.
9 July 2020
The close-knit computer engineering community has been rocked to its foundations by a scandal that has so far sparked three investigations. The facts to date:
An engineering PhD student dies of apparent suicide at the University of Florida in June 2019, leaving a note alleging research misconduct against his PhD supervisor.
Later that month, his supervisor, who denies the allegations, presents a paper they wrote together to their discipline’s most prestigious conference, the International Symposium on Computer Architecture (ISCA), held in Phoenix, Arizona.
In December 2019, an investigation into the co-written paper’s peer review process by the two computer engineering professional bodies that ran the conference, the Association for Computing Machinery (ACM) and the Institute of Electrical and Electronics Engineers (IEEE) concludes there was “no evidence of misconduct”.
In January 2020, screenshots claimed to be from the PhD student’s laptop, including alleged confidential peer review comments from the co-authored conference paper, are posted anonymously on the digital publishing platform Medium. The Medium article queries the findings of the peer review investigation, alleging that Chen’s laptop also contains hundreds of confidential draft papers by other authors submitted to the conference as well as their referee reports. It includes screenshots which appear to be lists of the papers.
An investigation by the University of Florida into the suicide-note allegations by the student, Huixiang Chen against his supervisor, Professor Tao Li, is ongoing. Ron Kozlowski, the attorney representing Chen’s supervisor, Tao Li, says: “We cannot comment except to say that Prof. Tao Li has cooperated fully with investigators and will continue to do so. Thorough investigations will exonerate him of any wrongdoing whatsoever in this tragic situation.”
In February, after publication of the Medium article, the ACM and IEEE launched a further investigation. The committee for that joint investigation consists of ethics and plagiarism experts as well as outside lawyers and others. Emily Reisbaum, an attorney on the committee, said in February that because the committee’s work is underway it would not comment on the matter. More recent requests for comment have gone unanswered.
Melissa Russell, executive director of the IEEE Computer Society, emailed members in February saying the society backed the investigation by the symposium’s two organizing bodies.
“The goal of this investigation will be to determine if policy or procedure violations occurred,” the email said.
A very big deal
For researchers working in the field of computer architecture — a branch of computer science and engineering — ISCA is the most prestigious and respected conference in the world, says Jose Renau, a computer engineer at the University of California, Santa Cruz, who was a member of the program committee, which was part of the reviewing process at ISCA 2019.
The discipline is unusual in that its researchers submit whole papers to the conference that go through rigorous peer review, instead of just submitting abstracts and titles, or conference posters or presentations, as happens for most other academic conferences. Renau says that most researchers only get a chance to present a few papers at the ISCA conference in their lifetime. “It’s a very big deal.”
The Medium article also had screenshots of lists of papers submitted to a related 2019 conference called the IEEE International Symposium on High-Performance Computer Architecture (HPCA).
Nature Index has not been able to verify the claim that papers by other authors submitted to the two conferences and their related referee reports are on Chen’s laptop, nor ascertain who posted the Medium article and screenshots. Even so it is clear that the confidentiality of the conferences’ paper submission process was compromised. Academics in the field believe an unprecedented breach of trust has occurred.
The Medium article was authored by a person or persons who called themselves “Huixiang Voice.” A related Twitter account featuring the same avatar bears the handle “Huixiang’s suicide and his struggle against academic misconduct”.
The situation is “profoundly disturbing at many levels”, says one senior computer scientist, who preferred to remain anonymous, speaking specifically about the confidential papers being circulated online. “A lot of people are very shaken by this.”
“I frankly don’t think that we thought that level of paranoia was necessary,” the anonymous computer scientist adds. “I’ve been in the field 20 years, I’ve been associate editor, I’ve been program chair, I would have never thought that someone would do this.”
It is “extremely unethical,” agrees Renau. He told Nature Index that the reviewing process of ISCA 2019 was split over two days. Half the papers were discussed each day with half the committee members present. Each study was refereed by four different researchers.
No single program committee member had access to all the submitted papers, Renau notes, so how they seemingly came together for the Medium post is a mystery. For Renau, it’s also concerning that the computer architecture community as a whole has been unusually quiet on the issue of a major ethical breach. “It gives a bad impression.”
An attempt to organize a boycott of ISCA 2020, which was originally scheduled for 29 May to 3 June 2020 in Valencia, Spain, but switched to online-only, was met with a strong response from the executive committee of the ACM’s Special Interest Group on Computer Architecture (SIGARCH).
In a post entitled “Remembering Huixiang Chen through our actions”, published on the anniversary of his death, the committee wrote: “In connection with this tragedy, and elsewhere in the broad computing community, there have been disturbing allegations of multiple violations of code of ethics and review processes in multiple conferences in computer architecture and across computing ... We reject calls to boycott our conferences and or calls that insinuate all conference organizers are complicit in large-scale collusion."
Renau suspects that either the system that handles the conference papers was hacked or more than one program committee member was involved with leaking the papers. Renau suspects it’s the former, noting that different people were in charge of the peer-review system at the ISCA and the earlier HPCA conference, while the software used was the same.
Eddie Kohler, a computer scientist at Harvard University, who built the software used by both conferences, said on Twitter in January that he thinks the papers got out as a result of being leaked rather than a hack. The ISCA papers appear to have been downloaded before the reviewing process began, he tweeted, while the HPCA papers seem to have been downloaded during a program committee meeting.
On 22 June, the ACM SIGARCH wrote that there will be some changes at ISCA conferences, including tweaks to the software used so that it “now tracks all downloads, a feature that can provide concrete evidence to ACM in reporting violations, and offers two-factor authentication for enhanced security”.
Earlier in June, the ACM leadership noted in an update that the investigation process “may be frustrating to those close to the case”, including the ISCA community. It offered an assurance that “we consider the potential wrongdoing as grave and will take appropriate actions based on any evidence exposed during the investigation”.
If you or someone you know is thinking about suicide, call the National Suicide Prevention Lifeline at 1-800-273-8255 or use the online Lifeline Chat, both available 24 hours a day, seven days a week, to those in the United States. In the UK and Ireland, Samaritans can be contacted on 116 123 or email firstname.lastname@example.org or email@example.com. In Australia, the crisis support service Lifeline can be contacted on 13 11 14. Other international helplines can be found at www.befrienders.org.