28 October 2026 · Sources checked 28 October · Resolved incident

The London outage: why a live voltage signal did not mean usable power

Google's October 27 report adds the missing control-logic explanation. This concerns the October 20 incident, not a new outage.

What the responsible source now says

The October 27 incident report describes voltage sensing upstream of an isolated breaker. Supply B remained energized there, although its breaker was open. The controllers interpreted that signal as a healthy supply and did not start the generators.

The report gives the main impact window as 06:00–12:08 UTC on October 20: 6 h 8 min. It separates the physical loss in europe-west2-b from wider regional degradation during failover.

Which summaries still reflect the preliminary report?

In our October 28 check, IncidentHub's October 24 snapshot retained 6 h 10 min and a pending final report. Deployflow's incident table also labelled the full report pending. Their snapshots should be read with their dates.

A reproducible counterexample

Our conceptual model below is original. It illustrates the reported false-healthy condition; it is not Google's PLC code, a hardware simulation or a configuration to deploy.

Duplicating a controller cannot resolve an identical ambiguity in both inputs. The discriminating question in this model is whether the signal describes voltage presence or an available delivery path.

const voltageOnly = voltageA || voltageB;
const pathAware = (voltageA && breakerAClosed)
               || (voltageB && breakerBClosed);

Inspect the original model and its independent check script. The check covers all sixteen Boolean states; this counterexample produces true / false. It does not validate real equipment or establish the vendor's exact implementation.

Built with AI assistance and source review. No accounts, uploads or paid model are needed. Reading measurement uses an ephemeral session; our checks are excluded. Privacy.