counter hit make
Trending

AI Tools

United's Ground Stop Is Not the Emergency. The Next 36 Hours Are.

4 min read

The tool is rarely the bottleneck. The process around it is.
The tool is rarely the bottleneck. The process around it is.

The outage is the least interesting part. United Airlines stopped departing flights on Nov. 16 over what it called a "widespread system outage," according to CNN. The New York Times reports the airline halted flights while it addressed the problem, and CNBC, citing the FAA, says the disruption affected departing flights. Every one of those sentences describes the first two hours of a story that is almost never finished when the systems come back up.

The question that decides whether this is a bad afternoon or a bad week is not what broke. It is what United can still do by hand.

The tell is in how the three headlines disagree

CNN calls it a "widespread system outage." The Times says United halted flights. CNBC, leaning on the FAA, narrows it to departing flights. That last detail is doing real work. An outage confined to departures points at dispatch, weight-and-balance, or check-in — something upstream of the airplane pushing back. "Widespread" points closer to the core reservation and passenger-service system, the thing that knows who is on which flight.

Airlines rarely say which one it was, and they will not for days. That gap between "widespread" and "departing flights" is the only real diagnostic we have right now.

The last three times

Southwest, December 2022. The trigger was winter weather. The failure was crew scheduling — the system that reassigns pilots and flight attendants when everything moves at once. It could not keep up, and the airline spent days rebuilding a schedule it could no longer compute. The lesson that has held ever since: the weather is rarely the cause. The cause is the system that has to absorb the weather.

Delta, July 2024. A faulty update from a security vendor took down Windows machines across the industry. Delta's own redundancy did not cover a vendor's blast radius, and its recovery tooling could not reconcile crews and aircraft fast enough to avoid days of cancellations. The lesson: your backup plan is only as good as your vendors' worst day.

The FAA, January 2023. The system that pushes safety notices to pilots failed, and the agency ordered a nationwide ground stop — the first of its kind since 9/11. The lesson there is institutional, and it is why the CNBC framing matters: the FAA is frequently the party that orders the stop. When a headline says the FAA was involved, that does not mean the government broke.

United itself has been here specifically before. Automation-related ground stops are a recurring event in its operations, not a black swan. That is the uncomfortable part of the precedent: the playbook is well-worn, which means the outcome depends on execution, not surprise.

What separates a bad afternoon from a bad week

Three things, and none of them are the outage.

Coupling. Is crew scheduling sitting in the same failure domain as the passenger-service system? If yes, the airline loses the ability to know who is on what and who can fly it at the same moment. If they are decoupled, the outage is annoying and survivable.

Manual fallback. Can dispatch run on paper and phones, and for how many hours before the workarounds themselves become the bottleneck? Every airline has this procedure. Almost none have tested it at full network scale recently.

Displaced metal and crews. Recovery moves in waves — hubs first, spokes later. That is why the second morning is the honest signal, not the first evening.

The AI angle nobody will ask about

Airlines have spent a decade stacking automation on top of core systems that predate the commercial internet: revenue management, crew tracking, dispatch, weight-and-balance. The industry's instinct is to add a layer, not to retire one, because the layer below is too expensive to replace and works well enough on a normal Tuesday.

Adding models on top of that stack does not make the base more resilient. It makes failures harder to diagnose, because the tools that tell you what is wrong are themselves downstream of what broke. So the question for United's technology leadership is not "was this a cyberattack." It is "what did the fallback actually look like, and when was it last exercised end to end?" If the answer is a tabletop exercise from a previous administration, you have your explanation.

If you are booked on United this week

  • Check your flight on the website or a third-party tracker, not the app. The app is a front end to the system that is reportedly broken.
  • Call. Airport desks and phone lines often ride on different systems than the consumer app, and a human with a working terminal can still rebook you when the app cannot.
  • Screenshot your booking and any delay notice. If you end up filing a claim, the timestamp matters.
  • If United cancels and you would rather not fly, ask for a refund to your original payment method rather than a travel credit. You generally have to ask.

Then watch the second morning. If United's first departure bank on Nov. 17 runs clean, this joins the long list of outages that were frightening at 8 a.m. and forgotten by Friday. If that bank runs late and thin, the story was never the outage. It was the recovery — and the recovery is the part airlines keep getting wrong.

Common questions

Was the United Airlines outage a cyberattack?

Nobody has said so. CNN, The New York Times and CNBC all describe a "widespread system outage" with no cause given. Most airline ground stops in the past decade trace back to internal automation failures rather than attacks, but until United says otherwise, unknown is unknown.

Are flights canceled or just delayed?

The Times reports United halted flights and CNBC reports departing flights were affected. A halt on departures almost always produces cancellations within hours, because aircraft and crews end up out of position for their next legs.

How long will it take United to recover?

Watch the first departure bank on the morning of Nov. 17. Airlines recover in waves, hubs first. A clean second morning means a bad afternoon; a thin one means the crew and aircraft tracking systems are the real problem.

Can I get my money back instead of a voucher?

If United cancels your flight and you choose not to travel, ask for a refund to your original payment method. Airlines frequently default to travel credit, so the request usually has to be explicit.

Related guides