Support
A prototype becomes an instrument when institutions use it
52 countries, nine capabilities, public data, open code. What it still needs is use, evidence and time. An institution can give all three.
The benchmark is free to read, free to quote and free to fork. Nothing on this page is a condition of using it.
Use it, and tell us where it broke
A measure nobody applies stays a hypothesis. The most useful thing an institution can do is put the benchmark against a decision it is already making.
- Read the thesis first. It states the claim under test and what would sink it.
- Then read the known limits. Some numbers say more about a missing dataset than about the country, and the project names which ones before anybody quotes them.
- Argue with a score at the challenge page. A dispute is published next to the number it targets, so the objection travels with the score.
Contribute data, evidence or a country layer
Every declared gap in the registry is a data-collection agenda item. Closing one is worth more than any commentary.
- A comparable series that covers at least two countries can become a scored indicator. CONTRIBUTING.md says how to propose one.
- A documented delivery, with a publisher behind it, becomes an evidence record. Records are never scored. They are what lets a capability with no dataset behind it still be read. The inclusion rule is short.
- A country layer is a second reading of one country, in that country's language and at that country's depth. Brasil is the first one, and the shape it uses works for any country whose institutions want to build the same thing.
- Everything else, including a bug, goes to the issue tracker. The code and the data live at the repository on GitHub.
Fund a piece of it
The work is modular on purpose, so a funder backs a named piece with a scope and a deliverable.
- A capability: one dimension, its indicator set, its gaps closed and its evidence corpus built out. Most research grants are already this size.
- A country layer: one country read in its own language, with its institution map and its subnational spread. Usually funded by a development bank, a school of government or a public foundation in that country.
- A source adapter: one publisher wired in and kept current, which raises confidence across every country at once.
- The expert panel: a reviewed run across the full country set, which is what turns session estimates into evidence.
Research grants, public-sector procurement, philanthropic funds and multilateral research budgets have all funded pieces of work shaped like these. If you hold one, or you know which window this fits, write to us and say which piece.
Keep the conversation going
The benchmark improves through argument, not through agreement.
If none of the above fits yet, the useful thing is still to talk. Tell us what your institution measures, what it cannot measure, and which decision it would want this to inform. One message reaches a person, and those conversations are where several of the gaps in the registry came from.