It seems to me that within any organization, there will be a particular group of people which can be depended upon to solve the majority of problems.
In turn, particular individuals within the group can be depended upon to solve a particular class of problems. Though it isn't necessarily predictable which person will yield the solution to any given problem.
I attribute this to distinctions in how each person thinks.
One of the most prevalent distinctions being between the theorists and the empiricists. Theorists live in their minds, on paper, and in the books. All things can be explained with math, mechanization, and sound judgment.
Empiricists don't count on things being predictable and want to look, to measure, and to mull over what they've seen.
Another important distinction seems to go between results oriented reasoning versus a more transcendental form.
Results oriented people seem to have more success with linking actions and reasoning back to the goals. However, this solution method tends to leave out the benefits of exploration in an effort to obtain rapid results.
Transcendental rational seems to have the opposite effect / evil. In a universe of diverse possibilities, the thinker becomes lost. Without the mooring offered by goals, it's easy to loose sight of the task.
With this in mind, you have a project manager who can be depended on to know realistic project expectations. You bring in your theorist and empiricist, who are both aware of the subject material. Then toss in a your imaginative person - who is also well experienced.
Then, give them time. Start with a full blown session with a nice meal, many presentations, and discussion. Then let the team work unfettered for a "cycle time," then give them a 1 week limit for presentations. Repeat this cycle about three times. Each time, let them work unfettered. Then apply a gentle pressure until presentation.
This method was used by the most skilled manager I've known - E Arrant