Stanke's first agile incompleteness theorem

I believe in agile. Who doesn't? But I'm also vexed by it. Who isn't? I was at a great presentation on agile product management this morning by Josh Seiden, and it helped me develop some thoughts around the friction that can arise between agile and business. I feel the root issue isn't a concern about process, but about results. Specifically, about defining when a project is "done."

Here's an imagined conversation:

Client/stakeholder: "Okay, so you're going to work on this project for a while, and when it's all done, you'll deploy it."

Technologist: "Actually, we'd like to do this the Agile way: we'll do a series of small Sprints, and deploy frequently, analyzing and iterating as we go."

Client: "Okay, sure. Whatever. You'll do your little sprints, and deploy frequently, until the project's all done."

Technologist: "Oh no. The project will never be done."

Client: "Say what now?"