Requirements Are Discovered, Not Collected

I received a request that sounded clear enough to begin.

Once I started working through it, several unanswered questions appeared.

What should the final output show?

What should happen when the expected record does not exist?

Was the goal a confirmation, a count, a link, a report, or a warning?

When I asked those questions, the request became much more specific. The new explanation included relationships, exceptions, statuses, and additional information that had not been part of the original request.

The work had not changed.

Our understanding of the work had.

I used to think good requirements were something a person could provide at the beginning if they were simply precise enough.

Now I think many requirements cannot be fully described until someone sees the shape of the solution and notices what is missing.

That does not mean the discovery process failed.

That is the discovery process.

I will not treat changing requirements as proof that the work was poorly planned. I will separate careless changes from useful discoveries and make room for the second kind.