In an article in the latest productmarketing.com magazine, Steve Johnson hit the nail on the head (as usual):
"One more point about market facts: We’re not interested in what features customers say they want as much as an articulation of problems they have. Henry Ford reminds us that if he had asked the market what they wanted, they would have asked for a faster horse. Customers will ask for more of the same while continuing to struggle with problems. Product managers need to identify those problems, not ask for the customers’ wish list of features.Documenting product requirements is nothing more than restating problems that prospective customers face in terms of the conditions that must hold to solve or avoid them.
Great product managers observe the problems and report them to Development—in the form of requirements."
I think it is harmfull if done too early!
The "right" path should be:
After identifying the problems (which lay behind the symptoms) and getting agreement on those underlying problems, you are ready to brainstorm with your customers on the direction of a solution to these problems.
When you got agreement on the direction of the solution (and not earlier than that moment) and the direction is "building a solution" you can brainstorm with your stakeholders on the features of the system that will be the solution to the problems.
That's the right time.
All features are held against the problems to be solved. Features that do not attribute to solving the problems should be left out of the system.