I especially agree with

I understand the mechanism explaining the effect very well.

When we're seriously in doubt, understanding mechanism can produce hypotheses to test. It's how we prune our search space.

In this case, we have two methods, (thanks for explaining what they were, by the way), both with plausible mechanisms to decrease mess. After even brief experimentation, it is obvious that method 1 has a specific flaw that prevents it from working well, and that method 2 works better because it suffers from no flaw as significant.

This last point is what is sometimes missed in (less careful application of) evo psych; plausible mechanisms are not necessarily well-understood mechanisms and probably can't be until they have been seen in action.