Uploaded November 2025 | Updated September 2026, 19 hours ago
For nearly a century, one scientific theory has shaped every radiation safety rule—and it may be wrong.
Subscribe to Abundance Institute on YouTube ► youtube.com/channel/UCP0Pf2ycSsqtEhgJZeT4N2w
Sign up for our newsletter ► https://email.abundance.institute/subscribe
In 1927, scientist Hermann Muller proposed what became the Linear No Threshold (LNT) model, which assumes that any dose of radiation could cause biological harm. His work earned a Nobel Prize in 1946, and became the foundation of global radiation safety standards: the idea that no level of radiation exposure is truly safe.
That belief led to ALARA (“as low as reasonably achievable”), a principle that leaves safety requirements to regulators’ discretion. Over time, it has produced inconsistent, overly cautious standards that make new nuclear plants costly to design and license with little or unclear safety benefits.
Nearly a century later, researchers are putting LNT to the test using machine learning and new data. In 2024, scientists from Columbia University and the Radiation Effects Research Foundation in Hiroshima, Japan, re-analyzed the gold-standard radiation-exposure dataset—the Life Span Study of A-bomb survivors—applying causal machine learning techniques to examine the causal relationship between dose and health risk.
Their findings upend decades of orthodoxy: at doses roughly below 50 millisieverts, they could not detect a causal link between radiation and adverse health effects. Similar preliminary analyses of nuclear-worker data show no measurable increase in risk at even higher exposures.
More on how we can fix the NRC and unlock nuclear power in the US: youtube.com/watch?v=Op3s5hpzgI0&list=PL2uT9dPo7lEhe1I_7t6N6wIL4lwEi1Npv&index=5
If regulation were aligned with this evidence, radiation protection could be right-sized: reducing costly, unnecessary radiation-control requirements, streamlining plant licensing, and helping advanced nuclear come online faster to meet rising power demand.
The study is currently under peer review. When the final version is published, we’ll update this description—but you can read the preprint version here: medrxiv.org/content/10.1101/2025.09.03.25335009v1
About Ray Rothrock:
Ray Rothrock is a Silicon Valley-based venture capitalist and former professional engineer who has championed efforts to modernize U.S. nuclear regulation. He co-founded the Nuclear Innovation Alliance, helped launch the film Pandora’s Promise, and has pushed for a shift from outdated prescriptive rules to flexible, risk-based licensing—arguing that regulatory reform is essential for the U.S. to lead in advanced nuclear innovation. You can find out more about Ray on his website: rayrothrock.com/.
Visit the Abundance Institute website to learn more ► https://abundance.institute/?utm_source=youtube&utm_medium=video&utm_campaign=youtube_description
For nearly a century, one scientific theory has shaped every radiation safety rule—and it may be wrong.
Subscribe to Abundance Institute on YouTube ► youtube.com/channel/UCP0Pf2ycSsqtEhgJZeT4N2w
Sign up for our newsletter ► https://email.abundance.institute/subscribe
In 1927, scientist Hermann Muller proposed what became the Linear No Threshold (LNT) model, which assumes that any dose of radiation could cause biological harm. His work earned a Nobel Prize in 1946, and became the foundation of global radiation safety standards: the idea that no level of radiation exposure is truly safe.
That belief led to ALARA (“as low as reasonably achievable”), a principle that leaves safety requirements to regulators’ discretion. Over time, it has produced inconsistent, overly cautious standards that make new nuclear plants costly to design and license with little or unclear safety benefits.
Nearly a century later, researchers are putting LNT to the test using machine learning and new data. In 2024, scientists from Columbia University and the Radiation Effects Research Foundation in Hiroshima, Japan, re-analyzed the gold-standard radiation-exposure dataset—the Life Span Study of A-bomb survivors—applying causal machine learning techniques to examine the causal relationship between dose and health risk.
Their findings upend decades of orthodoxy: at doses roughly below 50 millisieverts, they could not detect a causal link between radiation and adverse health effects. Similar preliminary analyses of nuclear-worker data show no measurable increase in risk at even higher exposures.
More on how we can fix the NRC and unlock nuclear power in the US: youtube.com/watch?v=Op3s5hpzgI0&list=PL2uT9dPo7lEhe1I_7t6N6wIL4lwEi1Npv&index=5
If regulation were aligned with this evidence, radiation protection could be right-sized: reducing costly, unnecessary radiation-control requirements, streamlining plant licensing, and helping advanced nuclear come online faster to meet rising power demand.
The study is currently under peer review. When the final version is published, we’ll update this description—but you can read the preprint version here: medrxiv.org/content/10.1101/2025.09.03.25335009v1
About Ray Rothrock:
Ray Rothrock is a Silicon Valley-based venture capitalist and former professional engineer who has championed efforts to modernize U.S. nuclear regulation. He co-founded the Nuclear Innovation Alliance, helped launch the film Pandora’s Promise, and has pushed for a shift from outdated prescriptive rules to flexible, risk-based licensing—arguing that regulatory reform is essential for the U.S. to lead in advanced nuclear innovation. You can find out more about Ray on his website: rayrothrock.com/.
Visit the Abundance Institute website to learn more ► https://abundance.institute/?utm_source=youtube&utm_medium=video&utm_campaign=youtube_description
