Estimate relatively, not absolutely

Humans are bad at saying how many hours something will take. But we’re pretty good at comparing two different somethings and deciding which is bigger. That’s why story points and t-shirt sizes are a thing. They let us estimate relatively.

I don’t know how many grams a cherry weighs, and I don’t know how much grams a potato weighs. But I’m pretty sure that a potato weighs at least 10x what a cherry weighs. So if a cherry is 1 story point, a potato is probably about 10. (Or if we’re doing fibonacci sequence story pointing, then a potato is 13.)

And maybe based on the last 6 iterations, we know that we’ve been completing around 30 points per iteration. That’s our average velocity. So based on that, we know we can probably bring two potatoes and a few cherries into the next iteration.

See what we did? We measured velocity and forecasted capacity using only relative estimates. So why would we bother trying to guess the weight of a cherry or a potato?

Therefore: why would we bother trying to guess the number of days an engineering task will take?

I also know that a cherry tastes better (to me anyway). So maybe we just bring in 30 cherries and declare potato bankruptcy.