Longer than you think, and less awful

Two wrong estimates keep this habit going, and they point in opposite directions.

The first is about time. On Sunday you tell yourself the proposal is "a couple of solid days". Where does that figure come from? Roger Buehler, Dale Griffin and Michael Ross studied the question in the 1990s and gave us the phrase "planning fallacy". People in their studies underestimated how long their own tasks would take, and the tasks ranged from academic work to everyday jobs. Asked to think aloud while estimating, they mostly built a story of how the work would go, step by step. Their own record with similar tasks hardly came into it. When the past did come up, they explained it away as special circumstances. That time there was the office move. The time before, they had flu.

Two details are useful. The error applied to people's own tasks. Observers estimating for somebody else erred in the other direction and guessed too long, partly because they paid more attention to how such things had gone before. We're optimists about ourselves only. And in one of the studies the bias disappeared, when people were told to recall similar past tasks and connect them to the estimate.

For someone who starts late, this matters more than it does for anyone else. If you start early, a bad estimate costs you some slack. If your whole plan is to begin at the last point from which the job can still be done, then the estimate is the plan. When it's optimistic, the overrun comes out of your sleep and the quality of the work.

So do what that study did. Think of the last three comparable pieces of work. Ignore what you planned for them. Ask how long they took, in hours, including the night. Most people who do this honestly find the number is bigger than "a couple of days", and that they've half known it all along.

The second estimate is about how bad the work will be. From outside, on a Sunday evening, the proposal feels like a wall. The dread is a forecast of how the work will feel, and I think it's a forecast made from the worst moment, the blank page with the judge awake. Twenty minutes inside the work, it's usually something much more ordinary. It's a series of small problems, some dull and a few interesting. You've noticed this on the last night, each time with surprise. If you've ever thought at one in the morning "why didn't I do this earlier, it's not even that bad", you've held the evidence in your hands. Then the memory that lasts is of dread and exhaustion. It gets filed under "proposals are awful", ready for next time.

Put the two together. You believe the work is worse than it is, so you don't want to begin. You believe it's shorter than it is, so you feel you don't have to. Each error makes waiting look reasonable.

You might say that you always make the deadline, so your estimate can't be far out. But you make it by cutting scope and sleep, which chapter 4 covered. Meeting the deadline shows that the deadline held. It says nothing about whether the estimate did.

Both errors are cured by the same thing, which is contact with the real job. Fifteen minutes inside it gives you a better sense of its size than any amount of thinking from outside, and a truer sense of how it feels. That brings us to the method.