Alex Wellerstein

@wellerstein.bsky.social

Nuclear historian. Prof@Stevens Institute of Technology. Visiting researcher@Nuclear Knowledges, Sciences Po, Paris. Senior Fellow@thebulletin.org. Author: THE MOST AWFUL RESPONSIBILITY (2025). Creator of NUKEMAP. Blogging @ https://doomsdaymachines.net.

When Edward Teller was chair of the early AEC Reactor Safeguards Committee, he was considered extremely conservative in his safety demands. His reasoning was that even one catastrophic accident could destroy all public enthusiasm for a technology he believed was important.

Carl Quintanilla@carlquintanilla.bsky.social · 2d ago

“.. AI companies are privately gaming out scenarios for a public and political revolt after a catastrophic AI event .. most likely a cyberattack, that shuts down access to financial services, internet connectivity, or even power and water.” @axios.com www.axios.com/2026/10/09/a...

They wrapped up their set with an instrumental rendering one of my all-time favorite children' songs, "The Song That Doesn't End." It's got such a wonderful trickster quality to it, a clever self-referential nature, and I've tended to think of it as a metaphor for the circle of life itself.

The Song That Doesn't End

YouTube video by G boz

youtube.com

Alex Wellerstein@wellerstein.bsky.social · 7d ago

There is what looks like a Catholic big-hair heavy metal band (big hair, leather, Jesus tattoos) loudly playing a brass band instrumental rendition of The Cure's "Close to Me" under my apartment window at the moment. Vive la France?

There is what looks like a Catholic big-hair heavy metal band (big hair, leather, Jesus tattoos) loudly playing a brass band instrumental rendition of The Cure's "Close to Me" under my apartment window at the moment. Vive la France?

I find all xenophobia and anti-immigration sentiment disgusting and offensive, but perhaps because I have spent so much of my life around Indian immigrants and Indian-Americans, the anti-Indian sentiment currently rising on the American right is just beyond the low expectations that I had for them.

As part of a system that was used to legally verify my identity online, a site asked me a series of "knowledge-based questions" — what is my SSN, birth date, former addresses, where I went to school, etc. — and then, this one... note that I already had to provide my DOB...

"Knowledge-Based Question. Using your date of birth, please select your astrological sun sign of the zodiac from the following choices."

There is a special place in hell for websites that have a phone number input field that is hyper-specific about what kinds of formatting it will accept for input, but they never tell one what the formatting requirements are. Parentheses? Spaces? Dashes? Country code? Sorry, you have to guess!

Quite a remarkable map, table, and diagram from a 1971 report on launch-on-warning, showing the implications of Soviet submarine launched ballistic missiles (SLBM) on detection, reaction, deliberation, and response time. All numbers are in SECONDS, which is what I find the most wild about it.

Map of the USA labeled "Soviet SLBM [Submarine Launched Ballistic Missile] Time of Flight to SAC [Strategic Air Command] Targets." Shows both "nominal" and "depressed" trajectories. Absolute best (north central USA) is between 580 and 850 seconds (9.6 – 14 min). Absolute worst (everywhere else) is between 320 and 650 seconds (5.3 – 10.8 min). The variations are based on where they assume the subs are launching from. Table: "Early Warning System Performance." Shows four different systems (BMEWS, OTHF, SLBM Net, and a not-yet-operational in 1971 Boost-Phase Detection Satellite) that indicates the time of detection after launch (varies between 55 and 150 seconds; BMEWS cannot detect launches as it is only over-the-horizon radar), the time that the US command and control system would be alerted after launch (ranges from 85 seconds for the newest system to 900 seconds for BMEWs), the probably of detection of a single launch (BMEWS is 99%, SLBM Net is 50%, none of the others have probabilities), probably of detection for a mass launch (all around 100%), and their false alarm rate (BMEWS has 1 per 7 years, OTHF is 1 per 6 months, the others are unknown). Diagram: "National Command-Warning to Response Cycle." Shows the consequences of a launch detection all the way through the US nuclear infrastructure, with "delay times" in seconds. Will be discussed in the next post in this thread.

So I find staying on top of e-mail IMPOSSIBLE. IMPOSSIBLE! How do you all manage it? How do you sort out the stuff you need to reply to or take action on in the future, and not lose it or forget about it? How do you keep e-mail from being 90% of your workday if that is not really your "job"?

Bild

Forget grandiose talk of curing cancer - this is what AI is actually doing to healthcare. In the words of one cardiologist: “it’s bots fighting bots, agents fighting agents, a horrible dystopic future nobody wants to live in.”

Bild

The Pope is visiting Paris tomorrow and will be literally across the street from me for part of it. I keep having these fantasies of running into him and having a banal Americans-in-Paris conversation. "Actually," I say to him, "if you tell them you're learning French, they're pretty encouraging."

From a strange conspiratorial Indian book on Google Books about how nuclear weapons don't exist (sigh) — but I thought these fruit analogies were fairly amusing. (The rest of the text, which is just that painful mixture of terrible confusion and disturbing overconfidence, is less amusing.)

"How much did Little Boy weigh? Everything about Little Boy is secret and mysterious. One of the project's great leaders, L. R. Groves, has told many thrilling stories about this bomb. But he was silent about how much Little Boy weighed. Some people think that two grapes' worth of uranium is enough for any nuclear reaction. Our letter writer does not think so. He thinks that Little Boy's uranium took only seven hundred grams in the reaction. That is not the weight of two grapes, that is the weight of one mango!"