The math and code of this simulation were engineered by Claude, an AI model made by Anthropic, from the concepts in Duke Johnson's book Better To Best and his related vision for eradicating extreme poverty while enriching cultures and supporting human flourishing. The model has not yet been reviewed by an independent economist; the full code is open for anyone to check, and expert collaborators are welcome.
An open agent-based model follows the same 500 simulated adults and their children, living in households as US adults do, for 20 years (or 40, with the Years switch below), once with Compassionism and once with no programme, in three environments. The figures below are averages over 500 paired runs; you can also run the model yourself.
node harness.js testbed 500 release ref,adv,st.
Dual-currency layer. Participants get monthly Basic Units (BU), spent on essentials at essential-service providers (ESPs), community-owned or private; expired BU are converted to primary currency (dollars) at merit-scaled rates through the Creative Collectives.
The community currency issued by CCO. Pegged 1:1 to primary currency, accepted for essentials by essential-service providers (ESPs), community-owned (PTF) or private, and expires monthly.
A business that sells essentials (groceries, utilities, health care, meals) and accepts BU. Community-owned ESPs are PTFs; privately owned ESPs pass what conversion adds to their BU customers as lower prices.
Networks of artists, makers and other contributors, part of CCO, that review members' work and manage the conversion of expired BU into primary currency at elevated rates.
Community-owned, not-for-profit businesses that accept BU. Modelled as a cost-reduction mechanism (12–16% lower costs), not a wealth-addition channel.
Community-owned housing; monthly payments build Acre Equity instead of paying market rent or mortgage interest.
Community equity built by PTH residents, appreciating ~4%/yr, accessible at exit subject to a tenure-based liquidity haircut.
Spatial governance layer organizing where PTF/PTH deploy; gates cooperative synergy effects behind a zone-participation threshold.
Digital platform administering CCO currency operations and democratic voting for PTH/PTF decisions.
Primary welfare metric: how many days of basic living an agent's accessible resources cover (cash + income + benefit buffers).
Share of income captured by contracts that build zero household wealth (rent, interest, fees). PTH payments have EDC = 0. The earlier engine's EDC figures are a design target proxy: a formula set to the BLEI paper's target bands, not measured from the model's flows (since v5.3 the model also measures it from rent and interest paid).
Monthly residual resources above the basic-living threshold; gates an agent's probability of advancing an octave level.
The answer in ten seconds
Four measures of poverty, three for adults and one for children, each an average over the paired runs. Use the strip above to change the environment, 20 or 40 years, and whether the big number shows the run with Compassionism or the run with no programme.
Since v5.3 the adults live in households with their children, start with the wealth the US survey shows for their kind of household, spend less in a year their income is short, and their BU buy the Hub’s full list. These change the no-programme run too, so the figures are not comparable with earlier releases (the findings explorer keeps each release apart).
Below 30 days of basic living
10.4%
with Compassionism, against 25.4% with no programme
Change: −14.9 points
In a typical year, the share of adults whose savings, pay and support would cover fewer than 30 days of basic living (the BLEI paper’s measure; in a household, its savings, pay and costs).
Below the cost of living
16.7%
with Compassionism, against 45.1% with no programme
Change: −28.3 points
In a typical year, the share of adults whose household’s income falls short of the household’s cost of a basic living (the MIT living-wage basket for its adults and children).
Too little wealth
36.0%
with Compassionism, against 30.5% with no programme
Worse: +5.5 points
At the end of the run, the share of adults with less than $25,000 of net wealth (what someone owns minus what they owe; in a couple, half of what the two own) in today’s money.
Children below the cost of living
29.4%
with Compassionism, against 72.6% with no programme
Change: −43.3 points
In a typical year, the share of children whose household’s income falls short of the household’s cost of a basic living (the MIT living-wage basket for its adults and children).
Reference environment, 20 years; averages over 500 paired runs of 500 simulated adults and their children, in households (release v5.3 data).
Go deeper: every result, chart by chart
Meet the adults
Each mark is one simulated adult in one run (the first of the paired runs), with Compassionism and with no programme: the same people, the same luck. Its shape and colour show the adult’s situation in the chosen year, the most serious that applies. Move the year, or choose an adult to follow their life. One run varies more than the averages above.
The adults of one run load when you reach this section.
Over time
The share of the same adults below each line, year by year, with Compassionism and with no programme. The line is the average over the runs; the shaded band is where the middle eight runs in ten fall, for the run chosen in the strip above.
Who gains
Where the programme moves people along the whole range of savings, and which groups gain or lose. A group that loses is named.
Households and children
The adults live in households as US adults do: single adults, single parents, and couples with and without children. A household is poor when its pooled income falls short of its own cost of living, childcare included, and each child brings a quarter of the adult allowance. Each row is a share of people, with Compassionism and with no programme.
What each part does
Each part in one card: what it does, what it adds to the result, the evidence behind how the model represents it, and its limit, with the readings that test it.
What each part adds loads with the release data.
The measure follows the choice in “How sure are we?”.
What it costs and who pays
Per adult a year, in today’s dollars, for the environment and horizon chosen above.
How sure are we?
Every reading added to test the answer (savings that keep up with prices, ageing, closer to US data, backing, rents, review errors, other ways to pay), as the change against no programme with its 95% interval. The vertical line is the main result; a mark to its left means the reading helps, to its right that it hurts. A reading that changes something outside the design (savings that keep up with prices, ageing, the US-data readings) is compared with no programme under the same rule.
| Reading | Below 30 days of basic living (your BLEI; design-neutral), change vs no programme | Below the cost of living | Too little wealth, year 20 | Prices, rise a year | Cost per adult a year |
|---|
The figures above for this environment and number of years, with the version, the command that reproduces them and the build fingerprint.
Run it yourself
A run here is one of the paired runs behind the figures above, so it varies more than their averages: the averages are the figures to quote. Its environment and years follow the strip above.
Explore the earlier engine The engine the page ran until v5.0 (v4.22, as coded), with its settings, presets and charts, on a page of its own. Its figures differ from the results above.