Estimations - Best, Expected and Worst

One of these posts I'll stop mentioning Dave Thomas, but until then...

Another thing that he mentioned was how to give estimates. Nobody should give a single point estimate. No risk factor is involved. But given a best case, expected case and worst case, it is quite easy to see what is risky and what isn't. With this in mind, we spent quite a few hours today giving high level estimates for our current work using these three points. (Work that we hadn't estimated but had planned to sometime this week. Isn't it nice how things come together sometimes...) I think everyone got something good out of it and our estimates feel a bit more accurate and considered.

To be honest, as a developer, I feel horrible when I miss an estimate, whether it is longer or shorter than expected. But with three point estimates, as long as we hit the range and are good about being close to our expected I don't think I'll mind missing our expected estimate (which is the only one we gave in the past).