Why thousands of fake scientific papers are flooding academic journalsPublished inThe Medium BlogSent as aNewsletter3 min readJust now-- Hello!Issue #265: auditioning for SNL, an engineers Hippocratic Oath, and waking up at 5 a.m.By Harris SockelBuilding on yesterdays theme of not taking what you read online at face value, I want to share a new report coauthored by two computer scientists and a former health reporter who, after a six month investigation, discovered a disturbing trend: thousands of fake academic papers, especially in cancer and Covid-19 research, flooding the system and blocking real science.Fake means written by AI (or at least with a lot of help from AI). In academia, more papers usually leads to more funding, so theres an unspoken incentive to churn out whatever will get past journal editors (aka publish or perish). This can have disastrous consequences. Not only do bogus papers mislead scientists trying to do real research, they can actually make people sick. One example: Research used to recommend ivermectin, an early Covid-19 drug, was later found to be fraudulent and not based in real clinical trials. One man spent nine days in the hospital after taking the pill.One of the engineers who coauthored the report created the Problematic Paper Screener, a tool that trawls millions of new scholarly reports looking for clues of deceit. One giveaway: awkward writing phrases that no one would ever say in conversation (crude information instead of raw data). If you see that, someone may have used AI to rewrite someone elses academic paper, subbing in verbose synonyms to evade plagiarism detection software.You cant just read an abstract and have any faith in it, says a cancer researcher quoted in the story whos been digging into this, I kind of assume everythings wrong.What about peer review, you ask? From everything Ive read, peer review kind of feels like the book blurb industry a system running on personal connections and favoritism. Most people who write a blurb (or a peer review) do so with some level of motivated reasoning. Ill blurb you, youll blurb me even if we dont love (or even read) each others books and the cycle continues. Also, ~17% of peer reviews are now written by ChatGPT.You can read the full report here. Im sharing it because, yes, its shocking, but its also a lesson in incentives. (Some) governments fund institutions purely based on metrics (# of papers published), and use quantity as a proxy for quality. Thats a bad way to measure value. But the root problem, I think, is this: Most people never read these academic papers. Theyre written for the writer, not for readers. A statistician quoted in the report sums it up well: We need less research, better research, research done for the right reasons.* Were also readingIn honor of Kendrick Lamars historic performance at the Superbowl, Im re-upping a rec from issue #80: Leila Renees list of Lamars writing tips, including this: Concision is power. You can cut sentences, paragraphs, even whole pages out of the story and be okay.Comedian Stacey Smith auditioned for SNL three times and was ghosted by producers, but somehow came out of it with a lot less performance anxiety: Those auditions taught me something way bigger than getting the gig ever could: how to live with uncertainty. (Also, she snapped a covert pic of Colin Jost and Michael Che, who were sitting in front of her on one of the audition days.)Software engineerings new Hippocratic Oath: I will remember that I do not write code just for fun, for business success, but for humans (Attila Vg) Some practical wisdomBryan Ye trained himself to wake up at 5 a.m. so he could start his day by pursuing personal projects, and learned the only way to start a new habit is to be honest with yourself about what youll lose when you do. (He lost about two hours every evening, because he now needs to be in bed by 9 p.m.)