Race on the racetrack, walk on ice

Our work involves a wide spectrum of activities, from early-stage RnD to certified flight systems. We are acutely aware of the differing demands and opportunities each of these environments presents us, and of which environment we are in. We are agile and adaptable in our thinking, and thus able to move from one environment to the next quickly without conflating these worlds.

Back to the universal principles

Under, just right, over.

SituationUnderJust RightOver
Early work treated as certifiedWraps an experiment in the full release process and learns nothing this month.Uses a light process, labels the work as early, and keeps it away from the certified path.Treats the experiment as if a failure would be harmless in the certified system.
A certified change treated as a sketchEdits a certified system with the habits of a prototype.Uses the checks that system requires, and does not borrow speed from the research bench.Adds research-grade ceremony on top of the certified checks and misses the real gate.
Two worlds in one weekUses whichever habit you used last, regardless of which world this task is in.Says which world this task is in before you start, and switches habits on purpose.Writes a new process for the switch and spends the week on the process.
A process from the wrong worldCopies the certified checklist into the lab, or the lab's shortcuts into the certified line.Keeps a short list of what differs, and follows the list for this task.Maintains two full process empires and makes people serve the process instead of the work.
Rigor that matches the riskApplies one rigor to every task because it is simpler to remember.Raises the rigor when the harm of being wrong goes up, and lowers it when you are still learning.Argues about the classification for longer than the task would have taken.
Conflating the worldsReports an experiment and a certified result in the same breath, as if they were the same kind of evidence.Labels which is which, every time, including in the summary a leader will skim.Separates them so aggressively that a lesson from the lab is not allowed to inform the product.

Sources and essays on this principle.

See something wrong or missing? Fix it in Facet.