Why this exists.

The problem

Ask a simple question, such as how a particular operator performs in a particular city at eight in the evening, and there is usually no independent answer. There is the operator’s own reporting, a regulator sample too small to be local, and a scattering of one-off speed tests taken by people who were already unhappy.

Meanwhile, hundreds of millions of devices sit on those connections doing nothing most of the time. Each one is already an instrument. The missing piece is not the measurement, it is a reason to take it, a way to trust it, and somebody to pay for it.

The approach

Devices run small bounded measurements. Readings are signed at the device, pass a published set of gates, and are aggregated into regional cells. Organisations that need the resulting picture pay for it, and that revenue is what pays the contributors who produced it. The coarse public layer stays open.

What this will not claim

• That it has coverage it does not have. A region with no readings is drawn as a region with no readings.
• That a reading is independent when it was funded by the operator being measured.
• That a single number describes a connection. Readings are published as distributions.
• That verification makes fraud impossible. It makes it expensive, and the refusal rate is published.

Where it stands

Early. There is no client, no measurement network, no dataset and no revenue. This site is the product being designed in the open, and every figure on it is generated for interface development and marked as such. The honest summary is that the interface is ahead of the network, which is the right way round for the questions that still need answering.