Scotty to Kirk, "We cannot get the shields back in less than an hour, Captain. The Klingon attack cracked our dilithium crystal, and there's antimatter leaking everywhere. Engineering bay is full of it." Captain James T Kirk is already aware of the situation. Whenever Scotty talks about schedule, he is always "full of it." Nevertheless, Kirk, the good manager that he is, simply informs engineering, "We haven't got that much time left, Scotty. Do your best." Kirk accepts that Scotty pads his numbers and that this characteristic is part of what makes Scotty such a great engineer.
Faced with uncertainty, Scotty uses his judgment to come up with a schedule so conservative that he's sure it can be beat. In my opinion, that's a good way to run the conceptual, creative phase of any project. Engineers need a lot of padding in this phase, and they don't need anyone looking over their shoulder asking them to justify why, if it's only a cracked dilithium crystal, it's going to take a whole hour. Scotty doesn't really know what needs fixing (neither would you if Klingons had just attacked your ship), so he gives a very conservative overestimate. Meanwhile, Kirk has to dodge photon torpedoes for an hour or so with nothing but impulse power and fading shields.
Kirk to management, "During the exploratory, architectural phase of any project, you must let your engineers pad their plans for schedule, cost, and performance. You must also ensure your business case still works if they run right up to the limit on all three. They might, but I bet not. When you let engineers pad their schedules and don't hassle them about it, they will work like madmen to make sure they don't go over the limit. Often, they beat it. That's the secret of how I manage great engineers like Scotty."
Once your engineers have established the feasibility and initial architecture for your project, the execution phase begins. Kirk knows that the execution phase is the time to squeeze the last drop out of the cost and schedule. By the beginning of the execution phase, your engineers have presumably already resolved any major uncertainties in your project plan. Each engineer knows what they need to do and how to do it. This is the time when a smart manager finds a detail-oriented project leader like Sulu to lead the design team. Sulu says things like, "If we don't change course, captain, we will crash in 5.3 minutes." Sulu knows this fact with certainty, because celestial navigation is a well-defined, straightforward process, similar in many ways to an execution activity. The exploratory phase isn't that way. Different people should be responsible for your project during each stage. Great architectural gurus do not always excel at managing details.
If you are a manager, how can you tell when your "Scotty" is giving you an accurate schedule? Basically, this situation occurs whenever you assign him a project he has done before. The real Scotty has fixed the Enterprise many times, so he just knows what it takes.
When you assign your Scotty to a new kind of project, you must give him time to build some proof-of-concept demonstrations. These demonstrations are not full product mockups. They are not something you show to your investors or use as road-show sales tools. Instead, they are a collection of critical fragments of the design, each built up just to a state showing whether a design will work, how long it will take to complete, and how much it will cost.
For example, when developing a new high-speed-bus architecture, your Scotty will want to look at measurements (or simulations) of the settling time on the major bus structures under all conditions. For a new chip-packaging technique, he will need to see signal propagation and crosstalk data obtained from the manufacturer, simulated using a 3-D field solver, or measured from an enlarged scale model.
Until your Scotty knows how all the pieces of a project fit together, he can't commit to a fixed schedule. During this phase, don't hassle him about detailed schedules, and don't promise performance numbers to your customers. Just focus on helping him obtain the proof-of-concept data he needs, accept his naturally conservative estimates, and keep dodging those photon torpedoes. Remember, you don't want to irritate Scotty while he has his hands around your dilithium crystals.