Accept that you can’t predict the BEST solution.

“The ones that demand that you as a project manager shall know what behind next rock will probably appreciate a horoscope more than reality.”

If things keep changing, the best solution will not be the best solution in the end of the project, if it even will solve the problem. Instead an empirical process must be taken in complex situations.  You have to evaluate a lot of possible solution, discussing pros and cons with different kinds of designs. A complex problem have multiple solutions some are just better than others, but you can’t predict them. Some solutions are more efficient and more productive than others just like when you solving Rubric’s cube.