Logo Lanfrica

mapomo-africa/methodology

Creator:
map
Host:
Versioned estimation protocols, validation procedures and accuracy notes # methodology **How Mapomo estimates campaign spending, versioned so that any figure can be reproduced.** Publishing the method is not a courtesy. An estimate whose method is private is an assertion, and an assertion about political money during an election is worth nothing and costs a great deal. Everything here is written to be challenged, and the challenge mechanism is a public issue. Version **0.1**, August 2026. Imperfect and dated on purpose: a published v0.1 that institutions can cite and criticise is worth more than a perfect document that arrives after the first deployment. ## Contents | Document | What it covers | |---|---| | `principles.md` | The commitments that constrain every protocol | | `protocols/outdoor-advertising.md` | Billboards, posters, painted walls | | `protocols/unit-cost-calibration.md` | How a price is established and kept honest | | `protocols/total-spend-estimation.md` | Aggregation, uncertainty, coverage, comparison with declared finance | | `validation.md` | How we check ourselves, and what we publish when we are wrong | | `CHANGELOG.md` | Every revision, with its effect on already published estimates | Protocols for broadcast monitoring, rally attendance, digital advertising and convoy pricing are in preparation and will be drafted with the Lab, since those are the areas where twenty-five practitioners know more than we do. ## How a protocol is versioned Every estimate pins the protocol version that produced it, in `provenance.methodologyRef`. That pin is what allows a figure published in December 2026 to be reproduced after the protocol has changed twice. Revisions carry a changelog entry stating what changed, why, and **what it does to estimates already published**. Silently correcting a number is not acceptable. Correcting it with a dated note is the entire point. ## Challenging any of this Open a methodology issue. Disagreement with a number belongs in public. Argument about who the number makes look good is out of sco …