Signals and Shared Understanding in the Apollo mission

The story

About two weeks before Apollo 11 launched in July 1969, Mission Control ran one of its last training simulations of the lunar landing. Partway through the descent, the computer began showing program alarms the team had never seen in training, and Steve Bales, the 26-year-old guidance officer responsible for the landing computer, called an abort. At the debrief, as flight director Gene Kranz later remembered it, the simulation supervisor, Dick Koos, told them:

You should not have aborted for those computer program alarms. What you should have done is taken a look at all of the function. Was the guidance still working? Was the navigation still working? Were you still firing your jets?

The team argued with Koos at first. Kranz then gave Bales the job of writing rules for every program alarm and told him he did not care how long it took. By Kranz’s account, Bales called him late that night to say, “Koos was right. We should not have aborted.” Bales and Jack Garman, a software specialist who supported him from the back room, brought in the people who had written the computer’s code and built its hardware. Those people said the alarms could not happen during a landing, but Kranz wanted every one worked through anyway. The team sorted the alarms into ones that were safe to continue through, ones that were not, and ones that needed a decision at the time. The 1202 alarm was in the last group. They wrote the list by hand and kept it at the console.

On July 20, 1969, with the lunar module about 33,500 feet above the Moon, Neil Armstrong reported “It’s a 1202” and asked Houston, “Give us a reading on the 1202 Program Alarm.” In the back room, Garman and his colleagues found the alarm on the list. The rule the team had written for it was, in Garman’s words, “if it doesn’t reoccur too often, we’re fine.” As Garman remembered it, the computer was recovering and nothing else appeared to be going wrong. Eleven seconds after Armstrong asked, Charlie Duke, the astronaut in Mission Control who spoke to the crew, told them, “We’re Go on that alarm.”

The alarms came back four more times during the descent, including a 1201, and Houston kept telling the crew they were Go. The lunar module touched down about seven minutes after the first alarm. The next day, shortly before the crew lifted off from the Moon, engineers traced the alarms to a radar switch setting that had been sending the computer data it did not need, which overloaded it.

How we see it through leadership economics

In the simulation, nobody at the console knew what each alarm meant for the choice Bales had to make, which was whether to keep landing or abort. The programmers knew what each alarm meant inside the computer, but no one had connected that to the decision. An alarm could have come from a check the programmers had left in for testing or from a computer that was failing. Bales had no way to tell which, so aborting was a reasonable choice.

We see this as a problem of Information. A signal is useful only when the person who receives it understands what information it was meant to carry, and on a team, everyone who acts on it needs to understand it the same way. Kranz’s order got the programmers, the guidance officer, the back room, and the flight director to agree, before the flight, on what each alarm meant and what to check when one appeared.

Because that work was done ahead of time, the team could spend the eleven seconds during the landing checking the rest of the data instead of working out what a 1202 was. The list did not cover the overload itself, which Garman said later “was not a situation that had ever been tested, ever, that I know of.” It also left a judgment to make, since the 1202 was acceptable only if it did not come back too often.

Teams have signals of their own, such as a status report that turns yellow or a deadline that slips, and the people who see them can take them to mean different things. Before one of those signals appears, it helps to check whether everyone who will act on it understands what it is meant to tell them.

A quote we’re thinking about

It is a familiar and significant saying that a problem well put is half-solved.

John Dewey, Logic: The Theory of Inquiry (1938)