ICAO’s Flight Recorder Specific Working Group met in the second week of September as part of the long amendment cycle on Annex 6. The group’s papers live behind a members-only portal. That secrecy is itself part of the story. The public argument is older than any single session is what must be captured, what must survive, what may be streamed, and who is allowed to listen. Crash-protected recorders now include flight data, cockpit voice, airborne image and data-link. Lightweight cousins exist for smaller types. Duration has been climbing from two-hour cockpit voice tapes toward twenty-five hours on new large aircraft. Underwater locators were stretched to ninety days after wreckage spent months on the ocean floor. Deployable recorders that punch out of a fuselage and float with an ELT are written into the books as an option. None of that ended the fight. It moved the fight onto cameras, bandwidth and old fleets.
Investigators want parameters the aircraft already computes and a view of the instruments when the voice channel is noise. They want a recorder that does not stay in a wreck at 4,000 metres. They want enough audio that a long-haul sector is not overwritten by the next two. What the aircraft generates is larger than what the crash box is certified to keep. Quick-access recorders and health-monitoring streams already leave the aeroplane every day. The gap is legal, not technical. Annex 6 and national labour law treat cockpit voice and any cockpit image as investigation material, not a management tape. Unions in Europe, India and North America have spent a generation drawing that line. An airborne image recorder that sees the captain’s hands is, to a safety board, a way to read a switch. To a pilot association it is a camera in a workplace. EASA already requires a crew-operated function that can obscure image and audio so normal replay cannot retrieve it. That compromise exists because the alternative was no camera at all.
Deployable recorders and real-time streaming look like the answer to another MH370. They collide with cost and false deployment. A new-build can take a deployable unit in the tail. A twenty-year-old A320 or 737 cannot take the same installation without a structural and wiring job that will lose the aircraft for a check. Streaming selected parameters through ACARS or a satcom pipe helps when the wreck is never found. It does not replace a crash-protected medium if the last minutes are a stall the airline never paid to downlink. Storage on the ground creates a second privacy fight: who holds the stream, for how long, and whether a prosecutor can ask. Retrofit is where the Standard dies. ICAO writes to the application for type certification after a date. The global fleet is mostly aeroplanes certified before that date. India’s narrowbodies, Indonesia’s islands fleet, Africa’s wet-leased metal, they will meet a twenty-five-hour CVR on the next delivered aircraft and keep a two-hour box on the rest until a State forces a retrofit that no lessor budgeted.
The working group’s useful product is not a slogan about more data. It is a parameter list that investigators can defend, an image class that labour law can live with, a deployable specification that does not fire on a hard landing and an honest note that older fleets will not be rebuilt to the new annex. Until those four things sit in the same amendment, aviation will keep arguing about the colour of the box while the argument that matters is which tails are allowed to fly without it.