How it relates to the 2023 report
What RP’s 2026 GCR fund model kept from Muñoz Morán (2023), and what it changed
Help improve this draftComment with Hypothes.is; flag @daaronr for a response.
Select any text, chart or number and annotate it with Hypothes.is (the tab at the top right) to flag errors, missing sources, or a clearer way to show something. David plans to respond, especially where a comment mentions @daaronr. Public Hypothes.is annotations can be read by anyone who opens this page, so don't post anything confidential.
How this was made: an AI agent (Claude, working for The Unjournal) rebuilt Rethink Priorities' GCR fund model from RP's public write-up, update log and Donor Compass data, and built these pages. David Reinstein set the task. Neither he nor RP has checked the numbers yet, and RP hasn't seen this page.
The rest of this site re-implements Muñoz Morán’s 2023 report for Rethink Priorities, The Value of Existential Risk Mitigation (interactive model, replication). RP’s 2026 GCR fund model, rebuilt in this section, lists the report as an influence alongside Tarsney (2020) and Thorstad (2022). It keeps the report’s core and adds much of what the report left for later.
| The 2023 report | RP’s 2026 GCR fund model | |
|---|---|---|
| The question | What is it worth to cut the risk of extinction a little, for a few years? | How many life-years does $1M to a specific fund buy? |
| Risk over time | Four paths; the main one is a time of perils: 20% per century for 100 years, then about 1% per century | A time of perils with uncertain timing: 20% (5% to 40%) in the next 100 years, a peak around year 17, high for about 90 years after the peak, then about 8 in a million a year |
| The action | Cuts all risk by a share f for P years | Cuts its own cause’s share of the risk (bio, nuclear or AI) by a small share per $1M, for about 11 years, starting after a few years |
| Can it backfire? | No: every analysis assumes the action lowers risk | 50% chance of no effect; 5% (bio, nuclear) or 15% (AI) chance of raising risk |
| Value of the future | Five paths, from constant to cubic or logistic growth | Logistic growth on Earth to 1.6 times today’s value; in 6% of futures, cubic growth as humanity settles the stars |
| Horizon | Up to 120,000 years | No end date; results split into six windows, the last 500+ years |
| Uncertainty | Fixed values; means only | Monte Carlo over every input (10 million draws since June 2026) |
| Smaller catastrophes | Not modeled | Three sizes, from 10 million deaths up, in the next 100 years |
| Units | Years of today’s world | Life-years |
| Where it goes | A research report | Per-fund numbers in Donor Compass and the Cross-Cause Fund, weighted by worldview and risk attitude |
What carries over
- The time of perils. In both, most of the value comes from raising the chance of reaching a long, safer era. In the report’s main case, 95% of the value arrives after year 500; in RP’s model, all but about one part in \(10^{14}\).
- A small, temporary, proportional cut in risk. RP’s funds work like the report’s action, with smaller effects per $1M and a delay before they start.
- Sensitivity to long-run risk. In both, the low risk after the time of perils sets how long the future lasts, and it drives the result more than anything else.
What changed, and why it matters
- Uncertainty and sign. The report’s evaluation by The Unjournal stressed that an action might backfire. RP’s model builds that in, and the Donor Compass page shows that the risk attitude applied to those harmful draws can decide whether a GCR fund gets money at all.
- Space. The report’s value paths stay bounded or grow polynomially; RP adds a 6% chance of cubic growth across the stars. Those futures carry essentially all of the expected value.
- No horizon. The report truncates at a horizon; RP doesn’t, and lets the 500+ window absorb everything after year 500. Donor Compass then puts one weight on that window.