EthniAfrica: before the states, there were peoples

Before nations and states, there were ethnic groups, peoples, kingdoms. That sentence is what started EthniAfrica, and it is what makes it something other than a directory: documenting the continent through its peoples rather than through the borders drawn over them.

This is not client work. It is ours: designed, built and maintained in-house, freely available on the web and exposed as an API. We are writing about it here because the interesting part is not the one you would expect.

Building a dictionary is no feat: records, relations, a search index, an API. The real work sits elsewhere, and the project states it plainly — reliable data on Africa is hard to come by.

So we cite. The UN World Population Prospects 2024/2025, the CIA World Factbook for ethnic distributions, World Bank development indicators, the UNESCO Institute for Statistics, and national statistics agencies, country by country.

Some sources are used with reservations, and the reservation is written down: the Joshua Project is flagged as one to handle with caution, given its religious orientation. Saying where a figure comes from is the minimum. Saying why you distrust it is the work.

Classifying peoples and languages means settling debates that are not settled. Plenty of databases get around this by showing a single classification and saying nothing about what it is worth. EthniAfrica does the opposite: every classification carries a status, and the status is visible.

Consensual, where contemporary scientific literature agrees and the debate is closed or marginal. Contested, where disagreement is live: disputed internal subdivisions, fuzzy boundaries with neighbouring families, competing documented hypotheses. There we keep the current classification but flag the controversy instead of hiding it.

Colonial heritage, for categories produced or frozen during the colonial period, generally by administrators, missionaries or linguists serving the administration. They are kept for historical traceability, explained for what they are, and self-designations come first.

Reconstructive, finally, where the classification rests on fragmentary sources: oral tradition, archaeology, genetics, glottochronology. That is owned as provisional and open to revision.

This is the part of the project we are proudest of, and none of it is technical. A database that shows its uncertainty is more useful than one that displays a confidence it does not have.

The dictionary covers the continent's 55 countries. You can enter it from wherever you like: by country, by people, by language family. Each entry carries the demographic information available and the links that tie it to the others — and it is that mesh, more than any single record, that makes kinship legible.

All of it is reachable through an API as well. A v2 is documented on the site, for anyone who would rather query the data than browse it.

An in-house project has no client to arbitrate for you. Nobody signs off a scope, nobody picks between two sources that contradict each other. You decide, you own it, and you live with the decision — including when the decision is to write down that you do not know.

That is exactly what we come for. On EthniAfrica the editorial doctrine cost more than the code, and it is the doctrine that makes the product worth anything. Useful thing to remember walking into client work: the hard question is almost never the technical one.

The dictionary is live at ethniafrica.com, freely accessible. It keeps being corrected and extended.