This reminds me of a battle between 'random walk' (no bias) and 'directed search' (preferred path) that we often face.

The same issue is in optimization (finding the best in a set of constraints). Often enough, random walk produces good solutions at lower cost than directed search. But people love 'preferred solutions' (because they love to see what they want to see?) even this may be more costly. Many researchers in optimization techniques praise 'random walk' but most use random walk in simulated (unreal) situations.

Preferred path is similar to accumulated small change, grafting or learning from history or experiences. Only limited options are evaluated and selected. Once committed, there are very few decisional issues to worry about, only few operational issues to make sure things get done (and ticked off the check list).

We can think about 'quantum' when all probable states are present, when a measurement is made (on one state -- all other states are ignored) or a decision is made, we no longer have many things (states) to deal with any more.

I would sum that up as "It is hard when we don't know what to do. It is easy when we make up our mind".