The connection was delayed by weather, so she arrived later than planned. The summer program coordinator met her at the airport. No emergency had occurred. Yet Lena understood that a system could be improved and still fail again. Change was not a promise that nothing would go wrong. It was a way to make errors easier to catch and less costly to the people affected.

She filed a report with Northstar. This time, the airline responded within three days. The accessibility office traced the failure to a software update that had hidden assistance requests when flights were rebooked. They corrected the display and added a warning when a linked request was not carried over. Lena asked whether they would test the update with passengers. The representative said they would invite the advisory group to a pilot session.

Lena returned to the youth advisory group for one meeting. She told them that the first fix had not prevented the second problem. The group was not discouraged. They helped design a checklist for rebooking: confirm the seat, assistance request, companion information, and connection plan after any itinerary change.

The checklist session took place in the same advisory room, with the same whiteboard, and it began, as good sessions do, with an argument about the first item. One student wanted the first item to be "Confirm your seat." Another wanted it to be "Confirm your name is spelled correctly." A third — the student whose wheelchair assistance had disappeared — wanted it to be "Get the name of the person who helps you." Lena listened, and then she suggested what became the compromise, and then the principle: the first item would be "Write down who you spoke to, and when," because every other confirmation, the group realized, depends on that one: a seat, confirmed to no one, at no time, is a wish, but a seat, confirmed to a named person, at a recorded time, is a record. The checklist grew, item by item, in the handwriting of five different students, which gave it the look of a document that belonged to no institution and therefore to everyone: confirm the seat; confirm the accommodation note, in words, not just in a code; confirm that the note survived any rebooking, by asking the agent to read it back; confirm the name of the person who confirmed it; and, finally, the item Lena insisted on, and the group adopted unanimously: "If something disappears, say so, in writing, to a second person." That last item was the whole checklist, really — the recognition that a single point of failure, whether it is a note, a sensor, or a tired technician's handwriting, is not a flaw in a person but a flaw in a design, and designs can be fixed, the way forms can be widened, the way thresholds can be set, the way a rover can be taught to pause. The group printed the checklist, simply, on one page, and gave it to the airline's advisory office, and to the school, and to anyone who asked, and Lena kept a copy in her notebook, beside the diagram of the pipe robot, because a checklist, like a robot, like a flight folder, is a machine for remembering — and remembering, done together, in a room, by people who have each lost something to a system, is the beginning of a system that loses less.

The airline adopted the checklist for a six-month trial. The staff who used it said it took less than a minute. Passengers said it reduced the need to repeat personal information. The trial became permanent.

The rebooking checklist trial revealed another problem: some agents treated a missing field as proof that no assistance had ever been requested. Lena recognized the same logic that had put her in conflict with Captain Cross. She proposed wording that prompted staff to look for an earlier record before closing the case. The technology team adopted a short alert explaining that the request might be linked to another itinerary version. The change was small enough to fit on one line, and that was part of its strength. It gave a hurried worker a reason to ask one more question.

At the summer program, Lena worked with students from across the country. They built a small rover to map a simulated disaster site. The robot had to identify safe routes and mark blocked passages. Lena liked the project because it required the machine to communicate uncertainty. If a sensor reading was unclear, the rover had to pause and request human review instead of guessing.