Why you can't finish anything you start

Updated 2026-09-12

The short answer

Starting costs almost nothing and finishing costs a lot, so anything that measures effort by things begun will keep producing more of them.

Each thing you start stays in the system until it is done. Four half-finished pieces of work are four items in progress, and by Little's Law the average time each one takes to complete is proportional to how many are open at once. Start a fifth and all five get slower — including the one you care about.

The feeling is that you lack follow-through. The mechanism is that you have too much open, and open things do not decay quietly. They accumulate.

Why starting is so much cheaper

Starting has no capacity check. Nothing stops you beginning something. There is no moment where a system says "you already have four of these open" — so the constraint has to come from you, every time, against a thing that feels productive.

Starting feels like progress and is recorded as progress. A new document, a first commit, a booked call. Each one produces something visible. The eighty percent that remains produces nothing visible until it is done.

Finishing is the expensive part of every task. The last stretch is where the awkward details are: the edge case, the bit you did not think through, the person you need an answer from. That is precisely the part with no momentum behind it, because the interesting problem was solved somewhere around the middle.

So the work that is easiest to do is the work that makes everything slower, and it feels good while you are doing it.

The compounding part

Every open item has a reload cost. Coming back to something means rebuilding where you were — what was pending, what you had already ruled out, why the obvious approach failed. None of that was written down, because while you were in it you did not need to.

With four open items you pay that reload every time you rotate between them, and that cost never appears on any list. It just quietly consumes the hours that were supposed to go to finishing.

Past a certain number the reloading dominates entirely. The formal name is thrashing: when what is actively needed exceeds what can be held, almost all effort goes into swapping and output collapses. Fully busy, nothing shipped. That is the shape of a week where you worked constantly and cannot say what got done.

What actually closes things

Finish before you start. This is the whole answer and it is unpopular because it means the interesting new thing waits. When something wants to begin, the question is which open item it displaces — not whether there is room, because there is always apparently room.

Make the last twenty percent its own task, with its own time. "Finish the report" hides the specific awkward thing that is actually blocking it. Naming that thing — send the draft to Priya, fix the numbers in section 3 — is often enough on its own, because the vagueness was the obstacle.

Count what is open, honestly, once. Not what is planned. What is started and unfinished, including the things you have stopped thinking of as active. This number is usually larger than people expect, and seeing it is most of the correction.

Check what is merely blocked. Some of the pile is not waiting on you. Something substantial held up by a short reply from somebody else is the cheapest possible thing to move, and it stays invisible until you look for it deliberately.

Cap the number, and let the cap bind. Three is a practical figure — small enough to force a choice, large enough to survive one interruption. The specific number matters less than that it is small and that you actually hold it.

The thing worth saying plainly

Nothing above is about willpower, and the failure described is not a character trait.

A queue with no limit on work in progress will accumulate unfinished items whether or not anything in it is trying hard. This happens in systems with no motivation to lack, and the fix in those systems is a cap on concurrent work — not encouragement.

If you cannot finish anything, the most likely explanation is that too many things are open. That is a property of how work is entering, and it is fixable by changing the entry rule rather than by resolving to be different.

Where to go next

• Little's Law, applied to a working day — the arithmetic in full, and what it does not claim
• The Top 3 method — a cap that binds
• What switching tasks actually costs — the reload you are paying for

Using this in Decide One

Decide One separates everything you have captured from the small number you committed to today, because those are different quantities and only one of them should be bounded. The captured list can be as long as it needs to be. The day cannot.

A line you did not finish is unfinished. It is not marked against you, and a day you did not open is not counted.

Open Decide One — free to use, with no account. Nothing you write leaves your device.