Highly effective charity discovered - 915 QALYs per dollar

By Will_Davison @ 2025-06-08T13:47 (+9)

I recently discovered that a group of people in Denmark has been taking 'the promise', which seems to run very close to something like the 'giving what we can' pledge. Its aim is to guarantee food, water, peace and shelter for everyone, forever.

To my understanding, nobody in the EA community has yet assessed this movement, so I created a cost effectiveness analysis. I was quite shocked at the initial results, which suggests 915 QALYs per dollar spent.

I would be glad to have your support in improving this CEA, and taking next steps in running initial trials.

What is the promise

The promise is simply a piece of text, which is read aloud to become a 'Promiser'. The person makes a promise that if any other Promiser asks them for help with obtaining food, water, peace (defined as freedom from physical violence) or shelter, they must either fulfill that request or bring in another Promiser to help.

Each Promiser creates their own 'jury' of five people that they know, whose ethics they admire. This helps them to stick to their promise and have a flexible method of solving moral conundrums.

The promise also has the condition that it comes before all other laws (which is less controversial than it sounds, given that it is restating a subsection the UN human rights convention), and if any Promiser is being punished for putting their promise first, you should either prevent the punishment or bring in more Promisers to help.

You can see the full text of the promise here.

Cost Effectiveness

I created this document to asses the cost effectiveness of a hypothetical charity that has paid employees spreading the promise. Example activities of the charity could be:

• Translating the promise
• Using networks to share the promise with individuals
• Improving the cost effectiveness analysis by monitoring success
• Generating media attention for the promise

Feel free to make a copy, trial different input parameters, and make any tweaks.

Key points from the CEA:

• The charity's first employee could continuously save lives for 0.04 USD per life, even with conservative estimates
• The promise has features that allow it to grow exponentially:
  • Low infrastructural requirements
  • Immediate benefit to those who join the promise
  • Benefit to Promisers of spreading the promise
• We need a trial run
  • There is a chance that the promise would not have exponential growth, and so would have a much smaller impact, if the defection rate (people dropping out of the promise) is greater than the spread rate. This should be tested in a trial
• Intervening early is key
  • creating the charity when there are only 100 promisers (estimated current number) versus 10,000 will increase cost effectiveness by 3x
  • Rapid action should of course be balanced with conducting trials and testing evidence
     

How can you help?

• Forecasting
  • What will the spread rate of the promise be over time?
• Test and improve the CEA:
  • Check for calculation errors
  • Refine estimates for QALYs lost through lack of clean water, physical violence, and homelessness
  • A full list of planned model improvements can be found at the bottom of the spreadsheet. Feel free to act on any of these
• Create a fancier CEA
  • Build an SIR-style model (similar to that used by infectious diseases)
  • Add scenario planning or uncertainties, e.g. through a monte carlo model
• Create and fund a charity
  • I suggest creating a small charity with two employees, to run a trial of spreading the promise and testing the CEA. Let's do this ASAP. Perhaps someone from Charity Entrepreneurship could take the lead?

Get in touch, and let's make a plan

P.S. this scheme could work well within EA

The promise would build a safety net from which EAs would be able to take high-impact actions that have personal risks, such as starting a charity or undertaking a research project without institutional funding. What do you think? Would you take the promise together with a group of EA friends?