How a few experts' judgment becomes a grant plan
The S-process takes the opinions of a handful of trusted experts and turns them into a single grant allocation — one anyone can reproduce from the inputs, dollar for dollar.
Below is a worked example you can poke at. Read it top to bottom: each step builds on the one before, and changing anything updates everything downstream.
There's money to give away
A shared pool, waiting to be split across a few organizations.
A few experts share the say
These are the recommenders. Each one's weight decides how much of the pool follows their judgment — weights are relative, so give someone twice the weight and twice as much money tracks their picks. Drag a slider and watch the plan further down shift.
Each expert rates every organization
A recommender doesn't just name favorites. For each org they say how valuable the next dollar is, and the point where more money stops helping. That's a value curve: high at first, falling as an org fills up. A flat red line means "I don't think this should be funded."
Whose curves do you want to see?
The grant plan
A dollar at a time, the pool flows to whichever org is valued most right now, switching whenever another pulls ahead, until the money runs out. Here's where it lands:
Which judgment moved the money
Because the money flows along each recommender's picks, every dollar traces back to whose conviction directed it.
How it got there
The mechanism never hands down a verdict — it streams money and changes course as curves fall. Every turning point, in order, reproducible from the inputs alone.
What one donor's gift would do
If a single donor added to the pool, their money would split across the plan in the same proportions.
Take it with you
The whole round — inputs, curves, weights, the allocation, and the replay above — is a single JSON file anyone can re-run and check.
Read the one-page spec this runs on — two implementations should agree to the dollar.