There is a specific feeling that arrives somewhere in the late stages of a project — not at the end, but near it — where the work becomes unpleasant in a way the early stages were not. The beginning had energy and possibility. The middle had problem-solving. The late stage has a different texture: relentless small decisions, no longer interesting, that have to be made correctly because the big decisions are already locked.

I have abandoned more projects in this stage than in any other.

What the Last Ten Percent Actually Is

The last ten percent is not a smaller version of the earlier work. It is cleanup, edge cases, and finishing — the parts that make the thing presentable rather than the parts that make it work.

In a writing project, it is the revision pass where you read the whole thing and notice that the third section undermines a claim you made in the first. In a software project, it is the error handling and the edge cases and the state you did not think about because you were focused on the happy path. In a physical project, it is the sanding and the painting and the part that will be visible and needs to look right.

None of this is intellectually interesting in the way that the design problems earlier in the project were. And none of it is optional, if the thing is going to be done rather than merely built.

Why It Is Hard

The early stages of a project have a feedback loop: you make a decision, the decision has a visible consequence, you learn from the consequence. The late stages have a different feedback loop: you fix a thing, the fix prevents a failure that does not now occur, you do not directly observe the benefit.

Fixing the error handling in a script does not feel like progress in the way that making the script first work felt like progress. But the script with broken error handling is not actually done.

This is also why taking notes during the early stages matters more than it might seem — by the time you are in the finishing stage, you need to be able to reconstruct the decisions you made earlier in order to know whether an edge case matters.

The Reframe That Works (for Me)

The late stage of a project is not the same work going slower. It is a different kind of work that has different satisfactions.

The satisfactions of finishing are:

  • The completeness of something that cannot surprise you anymore
  • The knowledge that the thing will work for someone who uses it outside the conditions you tested
  • The ability to set it down without it pulling at your attention

These are not as exciting as the satisfaction of a working prototype. They are quieter. But they are more durable.

A project that is mostly done takes up more ongoing mental space than a project that is completely done. The incompleteness requires maintenance. The finished thing does not.

That asymmetry, experienced enough times, makes finishing feel less like diminishing returns and more like the actual goal.