So what is the point of creating a specification if it is going to be wrong by definition? It is the process of creating the specification that is the most important aspect, or as Churchill said; “Plans are of little importance, but planning is essential.”
In new product development it is the process of scoping the market, listening to users, projecting technological trends, understanding business models, that is of most importance. Furthermore it is important that the product team goes through that process together, so that commercial and technical issues can be resolved creatively and progressively together.
So what does a specification, that is only transitory and whose validity is accepted, look like?
I believe that it should be a statement of needs based on specific evidence or assumptions. Forget all the stuff about dialogues and buttons. Leave that to the User Experience Designers who might have a clue what they are doing. Instead amass the evidence, negotiate the assumptions, then record that as a customer need.
As the product development progresses and you have the opportunity to show early results to target customers, the whole product team should be expecting to have to modify their plans. Listen to what the customers say, watch how their thinking develops about how they might use the product and how it would be of benefit to them, then modify the specification quoting that new evidence.
Approaching product specification this way there is a chance that your innovative new product will not be a first-to-market-flop, but at least have a fighting chance against the second generation products which have the benefit of seeing where you succeeded and failed before making their plans.