Tuesday, December 18, 2007

Building a Fort: Lessons in Software Estimation

Building a Fort: Lessons in Software Estimation is pretty entertaining and helpful too. I point out that another issue with his estimate may have been that he designed a fort that seems to be way more than his (or any) kids need.

He does point out that in this case the estimate didn't really matter, but he doesn't mention that doing an accurate estimate probably would have taken a full day (there I go, using round numbers to estimate). But if he really did a good survey of the land, inspected all his tools, got weight estimates of his materials, let alone drew a blueprint it would take time.

2 comments:

Roman said...

Regarding estimates, at place X we got into this "really need to produce accurate estimate and prioritize the work" fever. All the fuss ended up in a requirement that sounded like this "need XYZ fully functional and ready by March"....

Steve McConnell said...

Thanks for linking to my blog article. I don't see creation of the estimate as time consuming. For this project an hour would have been plenty. It's really the thought process that was missing. If had just looked at my materials list and even tried to jot down how many trips up and down the hill, for example, I would have added a defined amount of time for a necessary activity that I was essentially ignoring. This is parallel to software estimation errors -- a very common source of estimation error is neglecting to include all the work in your estimates.

And, for the record, my kids would not agree that the fort was more than they needed!