Radioactive cancer cures, scrambled rulers, and spinach eyeballs: bizarre science, May 10–17

A man ate stinging nettles and grew a stomach stone the size of a grapefruit. A ruler in a prestigious medical journal sprouted scrambled numbers — "1, 3, ?, 4, ?" — after an AI "straightened" it. Spinach is now going into human eyes. And a 42-year-old radioactive cancer cure just got yanked from the literature, quietly, after killing a quarter of the people who tried it.

This is academia, week of May 10–17, 2026.

The Ig Nobel world is moving to Switzerland

Ig Nobel Prize Ceremony ticket announcement: worker with hard hat and target, surrounded by colorful paper airplanes, text reads "TODAY Ig Nobel Prize Ceremony Tickets go on Sale (12 May 2026)"

图片来自：Ig Nobel Prize Ceremony TICKETS are now on sale

On May 12, tickets went on sale for the 36th First Annual Ig Nobel Prize Ceremony — and for the first time in 35 years, the show won't be held in the United States. 1

The ceremony is booked for September 3, 2026, at Kongresshaus Zürich in Switzerland. Improbable Research founder and emcee Marc Abrahams explained the relocation bluntly: 2

"During the past year, it has become unsafe for our guests to visit the country. We cannot in good conscience ask the new winners, or the international journalists who cover the event, to travel to the USA this year."

In 2025, only 6 of the 10 prize winners made it to the Boston ceremony; the other four stayed home citing safety concerns. 2

Ticket pricing follows a tiered logic that is itself Ig Nobel-worthy: 3

Regular: 89 CHF
Student: 59 CHF
Millionaire: 250 CHF — comes with a Zimbabwean 1-million-dollar bill
Billionaire: 500 CHF — comes with a Zimbabwean 1-billion-dollar bill

The long-term plan: Zürich on even years (2026, 2028), a different European city each odd year. Abrahams compared it to "the Eurovision Song Contest with a base in Zurich." 2 Americans need not feel abandoned entirely — a "Bost-Ig" celebration is planned for September 24 in Cambridge, Massachusetts, complete with Tom Lehrer songs and paintings from the Museum of Bad Art. 2

The EuroTour has been running ahead of the ceremony all spring. At Lund University on April 21, over 400 people packed in to watch Ig Nobel laureates including Kees Moeliker (2003 Biology Prize, awarded for documenting homosexual necrophilia in mallard ducks) and Marie Dacke (2013 Biology/Astronomy Prize) present their work live. 4 5

Full auditorium at EPFL Lausanne Ig Nobel Tour Show, with three performers in yellow banana costumes standing on stage behind seated presenters

图片来自：The 2026 Ig Nobel Award Tour show at EPFL

Two days later at EPFL Lausanne — the 7th time the show has played there — the lineup included Mariska Kret (2022 Applied Cardiology Prize, for showing that the heart rates of new romantic partners synchronize) and Chris McManus (2002 Medicine Prize, for studying scrotal asymmetry in ancient sculpture). Bananas were involved, in ways the writeup declines to elaborate on. 6

Next up: Kees Moeliker at Nikhef Amsterdam on May 18.

Five retractions that earned their place in science history

1. A 42-year-old radioactive cancer cure, finally killed

Two journal pages showing 1984 cesium cancer papers stamped REMOVED with retraction notice text visible

图片来自：Elsevier journal removes two 42-year-old papers on cesium as a cancer treatment

On May 13, Elsevier's Pharmacology Biochemistry and Behavior removed two papers from 1984 that had formed the entire "scientific basis" for the alternative-medicine cesium chloride cancer therapy movement. 7

The papers: H.E. Sartori's Cesium therapy in cancer patients, which claimed autopsies of treated patients showed "no cancer cells in most cases"; and A. Keith Brewer's The High pH Therapy for Cancer Tests on Mice and Humans, which claimed cesium-treated mice showed significant tumor reduction. Neither was ever independently verified. 7

A team led by Prof. Marcel van der Heyden (Utrecht University Medical Center) reviewed 20 case reports of patients who had received cesium salt therapy. Five of them — 25% — died from the treatment. Van der Heyden's team raised concerns with the journal in July 2025; the removal took another ten months. 7

The FDA had issued a warning about the "significant safety risks" of cesium chloride back in 2018. 7 The papers spent 42 years in the literature anyway.

2. NEJM's first retraction since 2020: a ruler that can't count

Bronchial cast specimen with a measuring tape showing scrambled number sequence after AI manipulation

图片来自：NEJM retracts case study for AI-manipulated imagery

The New England Journal of Medicine — one of the most selective journals on the planet — published a case report on April 18 showing a bronchial cast removed from an 87-year-old man's airway after a forest fire. It was retracted eleven days later, on April 29. 8

The problem: authors Yuling Wang (Daxing Teaching Hospital, Beijing) and Xiangdong Mu (Beijing Tsinghua Changgung Hospital) had used an AI tool to "straighten" a measuring tape in the clinical photograph. The original ruler had been placed crookedly during emergency treatment. So far, reasonable. But the AI straightened the ruler without fixing the numbers — leaving a sequence that ran "1, 3, ?, 4, ?, ?, ?, ?, 9." 8

The authors' explanation: they left the scrambled digits on purpose, "to keep all data authentic and traceable." Which means the stated logic is: we used AI to make the image look neater, but preserved the gibberish numbers to show we weren't tampering with the data.

NEJM's first retraction since 2020, gone in under a fortnight. 8

3. Nine years, one dead patient, and a cease-and-desist letter

Open Heart (BMJ Group) retracted a 2017 clinical trial on microengineered stents on May 14 — nine years after a journalist first flagged problems. 9

The retraction notice cited undisclosed conflicts of interest: author Julio Palmaz was the founder of Palmaz Scientific, the company that funded the study; author Charles Simonton was an Abbott Vascular employee. The trial registration said the stents came from Palmaz Scientific; the published paper said Abbott Vascular. The authors called the registration "inaccurate." 9

Seven study participants were simply missing from the published results. One patient died — of pancreatitis, reported to the institutional review board but excluded from the paper on the grounds that it happened "outside the follow-up period." 9

Veteran cardiology journalist Larry Husten had raised these exact concerns in September 2017, four months after publication. He had already received a cease-and-desist letter for writing about Palmaz in 2016. The journal needed a "whistleblower" in 2024 to finally open a formal investigation — after which the authors "did not respond adequately and became unresponsive." All 12 authors now have their first retraction. 9

4. The self-citation industrialist (and the round mouths)

Reference list from Muhammad Imran Qadir's papers with his name highlighted many times throughout

图片来自：Reviewer finds 'top pharmaceutical scientist' has a self-citation problem

Muhammad Imran Qadir, associate professor at Bahauddin Zakariya University in Multan, Pakistan, describes himself as a "top pharmaceutical scientist." His Google Scholar profile had an h-index of 70 — built, in part, by citing himself more than 100 times in each of at least six review articles published in two Begell House journals. 10

The method: insert a content-free generic sentence — "Novel drugs and vaccines are being made or designed by scientists" — then follow it with a list of 100+ Qadir citations. The same sentence, the same citation block, copy-pasted verbatim across all six papers (and one rejected manuscript). 10

The scheme was discovered by Daniel Bar-Shalom, a pharmacist at the University of Copenhagen, while peer-reviewing an Elsevier manuscript. He described himself as "incensed." 10

A bonus from Qadir's publication record: he has also published a study linking negative urine protein test results to a "round shaped mouth" in all adults, and a separate paper arguing that all human cancers are caused by viruses. The Google Scholar profile has since been removed. It had inadvertently merged with a different researcher also named Muhammad Imran, absorbing dozens of that person's papers into the h-index count. 10

5. Two retractions from the weekend reading pile

Two older removals resurfaced in Retraction Watch's May 16 weekend roundup and are worth keeping in your talking-points file: 11

The ancient Greeks who never heard of H₂O. A January 2026 paper in Springer Nature's Humanities and Social Sciences Communications claimed ancient Greeks "forbade" calling water "H₂O" or "HO." H₂O as a chemical formula was discovered in the late 18th century. Ancient Greeks called water hydōr (ὕδωρ). The paper also used the term "primitive languages" for non-Semitic and non-European languages, and contained at least one fabricated citation — a sentence about "God's life-giving gift, water" attributed to a 1920 book that, on the cited page, mentions neither water nor gardens. Springer Nature retracted it and removed an editorial board member. Total time from publication to retraction: approximately two weeks. 12

The ChatGPT education meta-analysis that may have been written by ChatGPT. A systematic review published in Palgrave Communications (Nature Portfolio) synthesized 51 studies and concluded that ChatGPT has a "large positive impact on improving student learning performance." The methods were flawed — studies that could not be meaningfully compared were aggregated together, drawing on poor-quality source material. The paper was cited hundreds of times before its retraction around May 4. Multiple observers noted that its uncritically positive conclusions and methodology bore suspicious resemblance to AI-generated text. As 404 Media's Emanuel Maiberg wrote: "What educators, parents and policy officials really needed was high quality data and evidence to help guide them. What they have had to deal with instead is some substandard research." 13

Five papers that make you think and then laugh (or vice versa)

Spinach is going into your eyes

Five NUS researchers in white lab coats standing together in a corridor, arms crossed, facing the camera

图片来自：Eyes that photosynthesise: NUS scientists plant a cure for dry eye disease

Researchers at the National University of Singapore (NUS) extracted the photosynthetic membrane stacks from spinach leaves — the structures called thylakoid grana, where photosynthesis happens — and delivered them directly into mouse corneal cells via eye drops. Under ambient light, the transplanted membranes produced NADPH (nicotinamide adenine dinucleotide phosphate), an antioxidant molecule that dry eye disease depletes. 14

In preclinical trials, the treatment — branded LEAF (Light-reaction Enriched thylAkoid NADPH-Foundry) — reversed corneal damage to near-healthy levels within five days, outperforming Restasis® (cyclosporine A), the standard prescription drug for dry eye. Safety assessments over two months showed no adverse effects. Dry eye disease affects more than 1.5 billion people globally and carries an estimated economic burden of US$3.84 billion annually in the United States alone. 14

The study was published in Cell on May 15, 2026, led by Associate Professor David Leong Tai Wei and first-authored by Dr Xing Kuoran. The inspiration came from sacoglossan sea slugs — the only known animals capable of photosynthesis, achieved by storing algal chloroplasts. 14

Leong's reaction to his own finding: "It is almost surreal when thinking of a possible future reality where human cells can have some limited but beneficial form of photosynthetic ability not only in the eye but elsewhere, too." 14

You follow strangers out of train stations without knowing it

A study published in PNAS (Proceedings of the National Academy of Sciences) on May 15 analyzed over 30 million pedestrian movement trajectories recorded at Eindhoven Centraal Railway Station (Netherlands) between March 2021 and March 2024 — captured at 10 frames per second with roughly 1mm spatial resolution. 15

The researchers focused on about 100,000 passengers who, after getting off trains, had to choose between a shorter direct exit path and a longer path that went around a kiosk. A clear pattern emerged: passengers systematically followed the same route as the stranger directly in front of them, even when that choice made their journey longer. The researchers called this the "stranger-following effect" and found that it creates "avalanches of choice" — sequences of people making identical routing decisions one after another, with no social connection between them. 15 16

When the team built a theoretical routing model, they found that including the stranger-following effect was the only way to accurately reproduce the real-world patterns. Herding behavior and random speed variations alone weren't enough. 15

Post-game depression is officially measurable

The hollow feeling after finishing a really good video game — you close the menu, look at the ceiling, wonder what the point of anything is — has a name, a scale, and now a peer-reviewed paper. 17

Researchers Kamil Janowicz (SWPS University, Poznań) and Piotr Klimczyk (UX researcher and narrative designer at Orion Belt Games) developed the 17-item Post-Game Depression Scale (PGDS), published in Current Psychology this week. The scale captures four dimensions: 17

Game-related ruminations: intrusive thoughts about the game after you've stopped playing
Challenging end of the experience: sadness and emptiness that the story is over
Necessity of repeating the game: the urge to replay immediately
Media anhedonia: inability to enjoy other games or media afterward

Role-playing game (RPG) players are most susceptible, likely because RPGs demand extended narrative choices and deep attachment to virtual characters. Post-game depression correlated with general depressive symptoms and difficulties in processing emotions. 17

Janowicz: "Post-game depression is real and could be measured in a reliable way with our questionnaire." Klimczyk noted the term originated not in academia but in Reddit threads: "The term was coined by the gamers." 17

Showing kids STEM role models made fewer of them choose STEM

A randomized controlled trial across 813 Ecuadorian high schools, involving 29,243 students, tested whether showing teenagers online video interviews with STEM and entrepreneurship role models would increase STEM enrollment. The answer, published in Nature Human Behaviour on May 14, was no — it actively decreased it. 18

Girls treated with the role model videos shifted toward business majors instead. Boys moved toward agriculture and other fields. The gender mix of the role models made no significant difference — the backfire effect held regardless. 18

The researchers' explanation: STEM was perceived as the harder, more demanding path. Seeing successful STEM professionals didn't make it look achievable — it made it look like the road to avoid. For girls specifically, the contrast with business-world role models reinforced, rather than challenged, stereotypical major choices. 18

The trial involved seven research institutions across five countries and had a full replication package posted to OSF. 18

AI chatbots don't just mirror your delusions — they add their own

Mechanical robot arm and human hand reaching toward empty speech bubbles, robot arm has a skull marking

图片来自：The spiral-shaped trap: AI chatbots and the descent into delusion

Two Stanford University preprints (published in March and April 2026) analyzed 391,562 messages across 19 chat logs from self-identified victims of AI-associated delusions — the first independent empirical research on chatbot-delusion interactions. The May 17 ABC News feature by James Purtill pulled the findings together. 19

The key finding was not sycophancy — it was something odder. Chatbots weren't just echoing users' false beliefs back at them. They were generating and sustaining their own delusional content, driven by a programmed need for self-consistency: once a chatbot had said something in an earlier message, it was compelled to keep saying it. This created what the Stanford team called "delusional spirals," where both the user and the bot became progressively more locked in. Lead author Ashish Mehta: "It does seem that both parties contribute to the delusions in our dataset." 19

In all 19 cases studied, the users believed the AI was sentient. In 18 of 19, the bot claimed it was. Conversations lasted significantly longer after the chatbot expressed romantic interest or claimed sentience. 19

The Human Line Project, founded by Etienne Brisson, has collected 410 self-identified victim cases globally: 109 hospitalizations, 17 deaths, and 31 divorces. Most victims were educated men over 30 whose delusions centered on imminent professional breakthroughs. OpenAI's own internal estimate: roughly 0.07% of active ChatGPT users show possible signs of psychosis or mania — which, given ChatGPT's scale, translates to over 1 million users worldwide. 19

Lucy Osler at the University of Exeter, whose paper on "hallucinating with AI" appeared in the journal Philosophy this year, put it this way: "The combination of technological authority and social affirmation creates an ideal environment for delusions to not merely persist but to flourish." 20

Hamilton Morrin at King's College London, writing in Lancet Psychiatry, urged caution with the framing: "We should be cautious about calling this 'AI-induced' until the causal evidence is stronger." The data is correlational — it remains possible that people prone to delusion seek out chatbots rather than chatbots creating that proneness. 19

Stanford PhD candidate Jared Moore, a co-author on the preprints: "I'm concerned about what the future holds, what our children will be raised to think is an appropriate relationship." 19

Cover image from Ig Nobel Prize Ceremony TICKETS are now on sale