I timed how long the “quick” tasks took to complete. The difference was stark.

For six weeks I jotted down my estimate prior to doing something, followed by the actual time spent on it. 94 tasks in all.

Any task that I estimated to take 15 minutes took me an average of 41 minutes. Any task that I estimated to take a day turned out to take an average of a day and a half. The scaling factor decreases as the task becomes larger – contrary to what I thought would happen.

Now when I consider this, I realize that the small tasks are the ones for which I never take the time to think beforehand. "Just change the copy." Forget that the copy is stored in three places and that one of those places is a hardcoded email template. Large tasks, in contrast, have already undergone some kind of decomposition, making their estimates well justified.

My solution to the problem is to refuse from giving any estimate lower than 30 minutes. If the task takes less than 30 minutes, I just do it rather than plan it.

Has anyone else ever measured this phenomenon? I am curious to know whether the small task inflation is a universal property or just my particular problem