One of my favorite approaches to solving a problem is to think about the meta-problem—and how to solve that larger problem. In some ways this is similar to root-cause analysis, because both solve a class of problems larger than the immediate problem given.
Recently, I discovered an abuse of the meta approach—a dark side that prevents people from getting any work done. Instead of solving the meta-problem, it creates meta-work. For example, let's say we believe we need to paint a bikeshed.
My approach is to find out why we need to paint the bikeshed because that will drive the decision about whom we hire and what we expect for results.
The dark side, however, asks us to immediately enumerate criteria for selecting paint colors—something that expert painters know how to do. The new work can then be argued ad infinitum because we can now argue both about which criteria have been selected, as well as how scores were chosen.
There is rarely an objective mechanism for selecting among preferences (that's what makes them preferences). The result is getting caught in a meta-loop in which no actual work is done.
Why Should I Care What Color the Bikeshed Is? at Bikeshed.com for the origin of the term "bikeshedding".
Parkinson's Law of Triviality at Wikipedia