Lesson Learned: who is best qualified to write a data specification?

[Roger L Costello <costello@mitre.org>]

Scenario: A data specification for aeronautical data needs to be written. Who should write it? Who best understands aeronautical data?

Intuitively, it seems that pilots would be supremely qualified for such an undertaking. After all, who knows aeronautical data better than pilots?

Alas, pilots may not be the best choice for writing the data specification. Let's see why.

Suppose that part of the data involves angles (perhaps the angle of an aircraft turn). The pilot says, "From my experience, the cockpit display shows the angle to 1 degree precision. Therefore the data should be to one degree precision." 

But what the pilot has failed to take into consideration is that the data might be arrived at via a series of calculations and those calculations might have roundoff errors. So if the specification just stipulates 1-degree precision, the data may end up being quite off (incorrect). Also, the pilot has failed to take into consideration the accuracy of the instruments that measure the angles.

Lesson Learned: When writing a data specification for a domain, a Subject Matter Expert (SME) in the domain is not necessarily the best person to write the data specification. At a minimum, the SME should be paired up with an engineer who understands things such as roundoff errors and instrument accuracy.

/Roger