Anders Sandberg

@idealfrontiers @DrPhiltill In practice one cannot do this kind of equating, and we might well have a risk bias saying loss of life is worse than slightly lowered life quality compared to optimum. But tail risks do not necessarily always win over small but pervasive improvements.

@idealfrontiers @DrPhiltill Actually... imagine the exercise studies improve quality adjusted life years by 0.01%; that gives you 800000 extra QALYs per year. Supervolcano risk is about 1/140,000 per year, starvation potential 5 billion people: expected risk 35,714 lives/y. Exercise actually wins.

Lava cooling is at the most practical here-and-now side of volcano engineering. The main ethical issue is cost effectiveness. nytimes.com/2026/03/15/wor…

@mechadense @DrPhiltill Yes, the term is heavily overloaded. It is weird to use the same term for modifying the entire atmosphere and local rock manipulation. Yet we found that the Oxford geoengineering principles (for the atmospheric case) actually worked well for volcanos.

@anderssandberg @DrPhiltill Huh. Geologic engineering at Earth scale is "geogeoengineering" then? I.e. toying around with biggest supervolcanoes with nonnegligible impact relative to their scale.

I like this overview. At least we know how to extend mice lifespan pretty well by now. Time to figure out how to fund proper trials in humans (the hard part is not the trial, it is the funding).

@mechadense @DrPhiltill In a magma melt there are multiple compounds mixed together. When it cools some starts crystalizing, pushing out volatiles into the remaining melt. This can bring it above the boiling point at that pressure, triggering an eruption. nature.com/articles/38551…

@anderssandberg @DrPhiltill What is "second boiling"?

@mechadense @DrPhiltill As we point out, people are already drilling into active volcanos for geothermal and soon hydrothermal mining. Lava redirection and lake gas release have a long history. These already have some ethical issues (e.g. causing risk to third parties).

@anderssandberg @DrPhiltill Technological and importantly economical feasability given the current (and forseeable) level of technology would be a question before the ethics of volcano control becomes pressingly relevant.

@shivon Largely small scenarios and vignettes, accompanied with diagrams. There is also commonly dialog between "me" and some interlocutor, where we explain or debate the problem.

When you think, what medium do you tend to think in? Would be very curious to hear how you’d describe the base unit(s) of your thoughts and how they feel to you. I assumed what happens in my head was similar to everyone else but have been surprised by how varied thought can be.

🎙️ HEET 2026 Speakers Include: @anderssandberg, Paul O’Hara, Vojin Rakić, Clare Shelley-Egan, Pieter Vermaas, Massimo Reichlin, Lílian Santos, Zeki Seskir, and more. 📅 April 16, 2026 📍 @CollegioCA 🔗 ieet.org/heet-2026/ #HEET2026

@DrPhiltill @budc1958 I think controlling mantle currents is surprisingly hard. In my grand futures work I consider terraforming and disassembling planets, but so far I have not hit on any good way of stopping mantle hotspots. Still, we likely have time to plan.

@budc1958 @anderssandberg It would require a vastly greater economy and physical capacity than we have today, but I don’t think there’s an effective upper limit on how large those things may become when we aren’t limited to a planet.

@DrPhiltill Oh yes! Do I have a non/profit for you! allfed.info

Interesting! I’ve also wondered if, as an easier problem, we could prepare a disaster response to the super volcano eruption to prevent loss of human life. It would require massive rapid evacuation and accurate prediction, and if the sky is darkened for many years afterward, it might require beamed energy from space for agriculture and extensive infrastructure for that. Areas that would be coated in ash, but not subject to the immediate violence of the explosion, might not need evacuation (where could they all be accommodated?) but would need plans to deal with the ash — security of water supply, etc. I think all those things might be doable.

6 weeks ago we released a report ecosystem detailing the state of brain emulation. We now added what is arguably the most important piece: A summary of the state of brain emulation anyone can read in less than 5 minutes and get the take-aways of thousands of hours of research. If you want to build intuitions on how close we are to running brains on computers, our at-a-glance summary is the place to start. For convenience I added the text in the thread below; please check out the full PDF for figures Accessibility was a key objective of our project and I think we delivered strongly with this one! Hopefully it pulls you deeper into our full report, the Asimov Press companion article, our public data repository, and an online guesstimator for predicting time and resources needed. Enjoy! brainemulation.mxschons.com

If you're an entrepreneurial scientist, you should definitely consider applying for one of the @AsteraInstitute 12-18 month fellowships. Current residents are working on mass-manufacturable space telescopes, reliable fusion energy TEAs, protein-based fibers, terraforming Mars, and more. They've got three tracks in the current call: Neuro/AI, life sciences, and “other.” Residents get: --Salary of between $125,000-250,000 annually (commensurate with experience) and benefits for the duration of your residency --Additional negotiated budget to cover project expenses (e.g. lab, infrastructure, compute, licenses, personnel), which can range from $0-$1.5M depending on project needs --Compute resources via Voltage Park, including on-demand access to 24,000 NVIDIA HGX H100s and access to a reserved fleet of B200s and GB300s --Dedicated workspace at our Emeryville hub with shared amenities and events --A community of like-minded technologists tackling bold challenges in intelligence and the life sciences Apply here (deadline is April 17) astera.org/residency/

@witchof0x20 I tried various celebrities. Then python. Then assembler. Then brainfuck...

Wait you can just substitute anything in the "to" parameter lmao. translate.kagi.com/?from=en&to=El…

("The canine has carried out a discursive capture of my schoolwork, integrating it into a consumer economy that subverts the disciplinary structures of the educational institution." - I think Sartre said it better!)

This is very amusing. Especially since I misspelled 'Dath Vader' and it still worked.