Interview
Strong longtermism, Pascal's mugging, and whether existing is good for us | Hilary Greaves (2020)
- Research from the Global Priorities Institute is expected to potentially shift views on practical global priorities in approximately two years.
- New work is anticipated to address whether it is better for a particular person to exist rather than not, alongside updated thinking on long-termism and moral uncertainty.
- A paper defining "axiological strong long-termism" (where best ex ante options involve the very long run) and "deontic strong long-termism" (where moral obligations align with best long-run expected effects) is being developed with Maul McCaskill.
- Strong long-termism is defined as highlighting a timescale of roughly one million years, contrasting with the 10-year focus often associated with general long-termist discussions.
- It is predicted that actions improving the far future could be many orders of magnitude better than short-term interventions, driven by an estimated average of $10^{15}$ future people in expectation.
- Specific risks identified include political lock-in mechanisms, path dependence in value systems, and the potential for artificial intelligence to embed extreme value systems within the next century or two.
- The argument for long-termism is expected to be strengthened by prioritarianism if future people are vastly worse off than present populations, though the simplest argument presupposes totalism.
- A significant positive rate of pure time preference could undermine long-termism by heavily damping future value in calculations, potentially setting value to zero beyond 100 years or causing exponential decay.
- The concept of "complex cluelessness" is introduced to describe situations where unforeseeable long-run consequences might cancel out in expected utility calculations, though arbitrary numbers may still influence decisions.
- Cluelessness is predicted to increase uncertainty spreads for both long-termist and short-termist interventions rather than systematically subtracting value or driving inaction.
- Three approaches to moral uncertainty are identified: dismissing the question, applying expected utility theory (the dominant predicted view), and a bargaining theory approach similar to maximizing expected choice worthiness.
- Pascal's Mugging is characterized as a paradoxical problem where discarding small probabilities with astronomical value is extremely difficult to formalize in a decision theory that works.
- Existence comparativism is defined as the view that it can be better or worse for a specific person to exist than not, leading toward totalist population axiology if being born with a good life is clearly better than non-existence.
- A heuristic for busy individuals is suggested to look for options robustly good across a broad range of moral theories, while acknowledging that specific cases may require gambling rather than insisting on robustness.
- The investigation into long-termist theses is noted to have revealed complex issues regarding population ethics and decision theory that remain unresolved despite recent paper writing.
- Future research is called for regarding affecting average very long-run future well-being, as this may prove intractable, although current efforts are considered extremely important and under-researched.