> No amount of planning or estimates will ever change that a feature will be done when it's done. I've been on teams where we got good enough at estimating tha

> No amount of planning or estimates will ever change that a feature will be done when it's done. I've been on teams where we got good enough at estimating that we were +/- a couple of hours each sprint on our burn down charts. Long tenure, good knowledge of the code, and a healthy amount of time spent on reducing technical debt all combine to make accurate estimations possible. Of course things can go awry in regards to partner team or 3rd party API integrations, those are always wild cards, but that is why you first do an exploratory task to figure out how hard integration will be (my current team calls these stories "spikes" in our workflow), which de-risks future estimates around the work. And it also makes more sense. "Spend this week learning how to use this API and figuring out its warts, then come back and give an estimate next week on how much work it'll be to get it up and running for real." A rather reasonable way to do things.