There needs to be a common understanding of iterative development, tradeoffs, solution design etc. It should be possible to get to pragmatic solutions without someone throwing weight. But it needs a good leader to get there.
Basically I do a poll of something like (a) confidence that the solution will fix the problem (b) estimated 80% confidence time range required. I do that for whatever options have been proposed, and then use the information to determine which option is the most promising. There's some principle that the average of many guesses is better than one, so I try to use that to make a better decision.