It’s about time

Time estimates, this little dance we do. I am not a fan of time estimates. Not at all. I avoid giving them if I can, because I have not come across many cases in my career where they have provided actual value. The only value that I feel come from time estimates is the deeper understanding of the feature at hand that comes from the discussion in the attempt to make the estimate. I often find the estimate itself rather useless.

Time estimates are in no way knowledge. They are guesses, at best. Any reasonable person, and most of us are, realizes this. We all know that there is a huge difference between what we guess and what we know. Still, ask anybody in a position where they demand estimates why they want them, and the answer will be along the lines of “well, we need to know how long time it’s going to take, when we can deliver”. Note the use of the phrase “we need to know”.

This means that the estimate, which at the time of creation everyone involved will agree is a guess, when moving towards the management side of the organization magically transition from a guess to a promise. And the ones that agreed that it was a guess and not a promise are the ones that will be held accountable for it in the end.
— “But you said it would take only three days!”
— “Um, no. We said it might take three days.”

My main issue with time estimates is not that they are not value-providing knowledge. My main issue is that they are often treated as such by those asking for them.

Still, more often than not I work in organizations where time estimates are required, an essential part of the development process. So, I have developed one habit that I try to introduce wherever I work, in order to soften the negative effects of them. I have come to realize that time estimates are also a form of communication. Apart from the obvious, they communicate a guess of the time required, they can also at the same time communicate another aspect; uncertainty.

My experience is that as the size of estimates go up, so does the uncertainty of them. When I communicate time estimates, if possible I do it in the unit where the estimate becomes a single digit. It’s four days, not 32 hours. 32 hours is anything between 31 and 33 hours. That is a pretty accurate figure. Four days is anything between, say, three and five days. That would translate into roughly 25–40 hours.

Saying “four days” or “32 hours” communicate very different levels of certainty. “32 hours” sounds so exact that it could almost be mistaken as a promise, which is exactly what I try to avoid.