THE SPECIFICATION

Before describing the role and creation of a specification, we need to introduce and explain a fairly technical term: a numbty is a person whose brain is totally numb. In this context, numb means "deprived of feeling or the power of unassisted activity"; in general, a numbty needs the stimulation of an electric cattle prod to even get to the right office in the morning. Communication with numbties is severely hampered by the fact that although they think they know what they mean (which they do not), they seldom actually say it, and they never write it down. And the main employment of numbties world-wide is in creating project specifications. You must know this - and protect your team accordingly.
A specification is the definition of your project: a statement of the problem, not the solution. Normally, the specification contains errors, ambiguities, misunderstandings and enough rope to hang you and your entire team.
Thus before you embark upon the the next six months of activity working on the wrong project,
you must assume that a numbty was the chief author of the specification you received and you must read, worry, revise and ensure that everyone concerned with the project (from originator, through the workers, to the end-customer) is working with the same understanding. The outcome of this deliberation should be a written definition of what is required, by when; and this must be agreed by all involved. There are no short-cuts to this; if you fail to spend the time initially, it will cost you far more later on.
The agreement upon a written specification has several benefits:
the clarity will reveal misunderstandings
the completeness will remove contradictory assumptions
the rigour of the analysis will expose technical and practical details which numbties
normally gloss over through ignorance or fear
the agreement forces all concerned to actually read and think about the details
The work on the specification can seen as the first stage of Quality Assurance since you are looking for and countering problems in the very foundation of the project - from this perspective the creation of the specification clearly merits a large investment of time.
The places to look for errors in a specification are:
the global context: numbties often focus too narrowly on the work of one team and fail to consider how it fits into the larger picture. Some of the work given to you may actually be undone or duplicated by others. Some of the proposed work may be incompatible with that of others; it might be just plain barmy in the larger context.
the interfaces: between your team and both its customers and suppliers, there are interfaces. At these points something gets transferred. Exactly what, how and when should be discussed and agreed from the very beginning. Never assume a common understanding, because you will be wrong. All it takes for your habitual understandings to evaporate is the arrival of one new member, in either of the teams. Define and agree your interfaces and maintain a friendly contact throughout the project.
time-scales: numbties always underestimate the time involved for work. If there are no time-scales in the specification, you can assume that one will be imposed upon you (which will be impossible). You must add realistic dates. The detail should include a precise understanding of the extent of any intermediate stages of the task, particularly those which have to be delivered.
external dependencies: your work may depend upon that of others. Make this very clear so that these people too will receive warning of your needs. Highlight the effect that problems with these would have upon your project so that everyone is quite clear about their importance. To be sure, contact these people yourself and ask if they are able to fulfil the assumptions in your specification.
resources: the numbty tends to ignore resources. The specification should identify the materials, equipment and manpower which are needed for the project. The agreement should include a commitment by your managers to allocate or to fund them. You should check that the actual numbers are practical and/or correct. If they are omitted, add them there is bound to be differences in their assumed values.
PROVIDING STRUCTURE
Having decide what the specification intends, your next problem is to decide what you and your team actually need to do, and how to do it. As a manager, you have to provide some form of framework both to plan and to communicate what needs doing. Without a structure, the work is a series of unrelated tasks which provides little sense of achievement and no feeling of advancement. If the team has no grasp of how individual tasks fit together towards an understood goal, then the work will seem pointless and they will feel only frustration. To take the planning forward, therefore, you need to turn the specification into a complete set of tasks with a linking structure. Fortunately, these two requirements are met at the same time since the derivation of such a structure is the simplest method of arriving at a list of tasks

references:
http://oldeee.see.ed.ac.uk/~gerard/Management/art8.html