Problems are not all alike
Morgan Jones’ The Thinker’s Toolkit separates problems according to their complexity.
The simplest problems—deterministic and direct problems—are those for which a single, exact answer exists. The answer is a fact and you only have to find the data (usually a Google search will do): how deep is the Danube river in Vienna, how many US president has there been, etc.
Next are the deterministic, computed problems.
An exact answer still exists, but you have to compute it: how many Euros were worth five dollars as of March 29, 2010; what’s the volume of a 20m pipe of inner diameter 10cm and outer diameter 15cm, etc.
Problems of the third kind become random but there is a finite number of solutions; i.e., you know that the answer is part of a group, and you can identify exactly all elements of that group. Examples of such problems include: what will be the outcome of throwing a dice, or will our company’s revenue exceed $5m next fiscal year.
Problems of the final type are random and continuous. There, we can only give ranges of where the solution lays, but cannot point to an exact answer: by how much will Apple Inc.’s share fluctuate after the introduction of the iPad, by how much will the earth’s temperature have risen as a result of human activity in January 2050, etc.
For deterministic problems, facts are crucial: there is an exact answer and either you know/find it or you don’t. As you move into random problems, however, the role of expertise grows. You now have to rely on past experience, intuition, “soft information”, etc. to get to the answer.
Adapt your level of confidence to the type of problem
Just as a mountain climber only trusts his rope in certain conditions (after having checked it before the climb, knowing who assured it, etc.), you shouldn’t misplace your trust in your solutions.
As the role of facts diminishes, the role of expertise rises and so does the probability of error in your solution process, so prepare to lower your confidence level.
Be honest with yourself and with your boss/client: differentiate those recommendations that come straight from hard facts from those that result from an analysis that required significant amounts of guessing.