Too Early, Too Late

I recently heard someone explain that project leaders are being held accountable to the cost estimates that are produced at "Milestone A" (i.e. VERY early in the project). This was considered a Bad Thing, because MS A is so very early in the project timeline.

Yes, there's something silly about setting a cost estimate in stone before very much is known about the project. But at the same time, perhaps the problem isn't necessarily that we're doing our cost estimates so early. Maybe the problem is that we're delivering the system so late! Go ahead and do your cost estimate at MS A. If you then deliver the system 1 or 2 years later, you'll probably be fine. A good cost estimate can probably survive the passage of that much time.

Sure, if it's a 10 or 15 or 20 year project, that initial cost estimate is bound to break down. But if the distance between estimate and delivery is too long, delaying the estimate isn't the only way to make things better...