How it was built
Between 2024 and 2026, in Bangladesh, people who were not paid salaries built a measurement instrument, tested it on a real population, reached 2,259 verified participants across 211 constituencies, and ended by founding an independent society that now owns the work. No institution funded any of it.
| Period | 2024 to October 2026 |
| People | Roughly thirty in sustained roles; 2,259 verified participants in the wider network |
| Reach | 211 constituencies across eight divisions |
| Salaries paid | None. Honorariums only, and only sometimes |
| Institutional funding | None, at any point |
| Costs met by | Individual contribution |
As they held their roles when the work was done. The measurement they produced is on Evidence.
Among many. Alumni of the programmes returned at every stage, and several of those named elsewhere on this page came back the same way.
2,259 verified participants across 211 constituencies and eight divisions, and the 86 who gave the consented responses the pilot reports.
Four.
Twenty-five questions across five domains, with a confidence measure that separates a settled judgment from a contested one. Tested on 86 consented responses in April 2026. It produced differentiated, interpretable results rather than noise, which is the only thing a pilot can establish.
Overall Fairness Index 1.74 out of 5.00, every domain below 2.00. And within it, the Q18 result: the strongest agreement anywhere in the instrument concerns education quality, which maps to a constitutional provision that is both judicially unenforceable and unamendable. Legal Injustice measured rather than argued.
Published in full — the instrument, the nine formulae, the versioning rules and the formation kit. Free, unlicensed, and reproducible by anyone holding raw responses. None of it had to be negotiated with a funder, because there was no funder.
On 1 November 2026, members subscribed a memorandum and founded CARO Bangladesh Society — governed by its own members, electing its own officers, able to remove anyone including the people who founded it. Most of the people named above are now members of it rather than staff of CARO. That was the point from the beginning, and it is the only output that can outlast everyone who produced it.
The society is weeks old and its registration is incomplete. The test is not what was built by 2026 but whether a local unit is still taking readings in month thirty, when nobody is checking. That answer is years away, and it will be published either way.