Method

How a vote becomes public

Local proof of concept for one area group. Not sanctioned by the Fire Brigades Union nationally. Not an official FBU ballot. For this trial the operator holds a dummy list of made-up membership numbers issued to testers. Favourite fruit and favourite colour are the test questions until the area trusts the count. Results are a demo of the machine, not union policy.

1. 01

   A dummy roll for the trial

   Testers are given a made-up membership number and last name. Nobody votes until they match a row on that list. Opening a poll freezes a snapshot of the roll for turnout. This is not the real area membership list. Next phase, if the area wants it: number + last name + email OTP.

2. 02

   A trial check, not identity

   The tester sends number and last name. Spaces and hyphens are ignored. Repeated failures lock the number and flag officials. This is not proof of identity. If the bot rejects you, stop and ask for a fresh test pair. Do not guess.

3. 03

   One test account, one vote

   After a match, that Telegram or web session is bound. Votes are counted per membership number, not per Telegram account. Names never appear with totals.

4. 04

   FBU Voice posts the poll

   Officials write a balanced brief, then open. An AI digest restates both sides in equal weight — it does not tell you how to vote. The bot posts buttons after /bind. Unverified taps are rejected.

5. 05

   Named cases, full names

   Members can publish a case on the poll page. Full first name and last name are required and shown. No anonymous posts. Officials can hide abuse, not silently edit substance.

6. 06

   Results publish themselves

   Charts, turnout, rejected attempts and a public audit list (no personal data) appear on crewvoice.uk as votes land.