Epistemic status: watched a YouTube video at 2x speed and now I'm smarter than everyone who ever worked at GPI (RIP)
Kurzgesagt recently released a video called "How a Future Civilisation Will Become Immortal", explaining how the last civilisation alive could survive the heat death of the universe. The idea, based on a Freeman Dyson paper, is that digital minds can progressively cool their brains, think slower, and hibernate through increasingly long cosmic nights. Each cooling cycle reduces energy costs so dramatically that a finite battery could sustain conscious experience for the equivalent of two trillion trillion trillion subjective years. Possibly forever. The maths checks out. The vibes are immaculate.
This has chilling implications for longtermism.
The standard longtermist case assumes a merely very large future: perhaps 1023 future people across our light cone. Organisations like 80,000 Hours use figures like this to argue that reducing existential risk is overwhelmingly important. But they've been working with a toy model. If Dyson's Cold Thoughts hold, the number of subjective life-years available to future civilisations isn't just astronomically large, it's potentially unbounded. The expected value of the far future isn't 1023. It's infinity.
This changes every cost-benefit analysis ever conducted. Under standard longtermist reasoning, spending £5 on a coffee is already justified if it makes you 0.000000000000000000001% more productive at reducing x-risk, because the future is so large. But if the future is infinite, then any action with any nonzero probability of marginally reducing existential risk has infinite expected value. Buying a £4 oat latte? Infinitely justified: you need to stay alert to write that crucial forum post. Upgrading to business class? Infinite EV: you'll arrive better rested for the biosecurity conference. A second monitor? The productivity gains compound across literally infinite subjective experience-years. A third monitor? Now you're just being selfish.
