# A fifty-node federated pilot

ID: node-pilot · experiment · proposal · updated 2026-09-07
Tags: deployment, experiment, federation

The experiment that would move the thesis from argument to evidence: fifty installed nodes, one utility or data-centre timing signal, one season, and a published record of delivered flexibility, comfort and compensation.

## Design
Fifty Airform nodes in buildings with consenting occupants. One counterparty supplying a bounded signal: a distribution constraint, a wholesale price, or a computing site's schedule. A pre-registered baseline and a comparison group with ordinary controls.

## What is measured
Delivered kilowatt-hours per request, at the meter; recovery demand; usable battery share by season; override count and latency; occupant-reported comfort; compensation paid and to whom; communication failures and the safe-mode behaviour that followed.

## What would count as failure
Delivered flexibility below the promised floor in more than a tenth of requests; comfort deviations outside the agreed envelope; or a compensation distribution that participants, surveyed afterwards, consider unfair.

## Why fifty
Large enough that a location analysis is possible and a fleet promise means something; small enough that a single team can watch every node.

## Counterargument
The benefit may disappear when tested against a simpler alternative or under adverse operating conditions.

## Next test
Name the counterparty, the site owner and the participant advocate; register the baseline before the first dispatch.

## Sources
- none

## Links
- tests → federation
- tests → dispatch-without-discomfort
- tests → ai-load-question