Methodology

How we turn blackouts into data

Everything here is public. If you can read this page you can check our work.

1. The Power Watcher network

Each monitored area has three independent Power Watchers who live in different parts of that area. They do not know each other's reports. They report only what happens in their own home: power off, power back.

2. Confirming an outage

One report alone is never published as an outage. When a second Power Watcher in the same area reports the same thing within the matching tolerance, the outage is confirmed. A single report is kept as "reports do not yet agree" (orange).

3. Combining timestamps

We never add durations from different Watchers together. We take the median of their timestamps: the middle value. This protects the record from one phone with a slow clock or one person who reported late.

4. Stabilization

Power often flickers back on and off. If the power fails again within the stabilization period, we treat it as the same outage and continue counting, rather than recording two short outages.

5. The GERI reliability index

The Gambia Electricity Reliability Index scores each area from 0 to 100 using four factors: how much of the time power was available, how often it failed, how long each outage lasted, and a penalty for very long outages. The weights are listed below and can be checked by anyone.

6. What we never do

  • We do not estimate outages for areas we do not monitor.
  • We do not publish individual Power Watcher reports or identities.
  • We do not present estimates as measurements. Estimates are always labelled.
  • We do not take political sides, and we correct our own errors in public.

Live parameters

The settings we run on