In the assessment and judgment of actions, particularly those that involve imperfect information such as in politics, we need to be clear about what it is that we mean when we say someone is right or wrong to do something.
In that regard, there are three separate axes of goodness and one of uncertain goodness, which are distinguished by the information necessary to achieve them:
- Omniscient optimal
- Correct (in the situation/under constraints)
- Justified
- Reasonable
Omniscient optimal, is what you would choose if you were God, and knew what card was coming, what individuals are thinking and will decide, and have every other answer to questions about the actual physical future. Sometimes we say “turned out to be correct/the right decision” but that is a sloppy way of thinking, since the evaluation of the average case of a decision-making approach involves considering both the situations in which the strategy would make the omniscient optimal outcome, and something short of it. While those poor outcomes do impact the evaluation, as statistically single instances, they do not necessarily change the correctness of the decision-making approach, and therefore not the assessed correctness of the specific decision made (as one of a large set of decisions) made under that approach.
Correct, is that you made the decision with:
- Logically coherent reasoning with a sufficient model of the situation to identify the best outcomes;
- Having gathered the amount of information either that was available, or was economically feasible to gather;
- Considered the value systems that are most appropriate to the utility of the agents on behalf of whom the decision was being made.
- And then chose the action that truly optimizes the value over all such instances of this particular situation; that is, there is no flaw a reviewer could identify that clearly would make another action better. (There could be multiple correct decisions within the precision of the model(s).)
Correctness has a structural distinction with omniscient optimal, being that in omniscient optimal, you choose the action only for this one situation/with no regard for other situations in the class.
As the space of all real life decisions includes decisions based on imperfect information, we cannot consider that omniscient optimal is attainable, and therefore that cannot be the standard by which we judge decisions, since we have to judge the correctness of our decisions before we make them.
Justified, is that you made the decision, having received enough information to cross a specified decision threshold that is considered to be a trigger for action. In that regard, you could consider it as a related concept to “honor” in that there is not one universal honor code, but rather a series of standards under which people evaluate your actions as honorable (or dishonorable). All correct decisions can be justified under the respective value systems (assuming those were chosen correctly), but not all justified decisions are most correct; you could be leaving something on the table.
Reasonable, is that you got some sort of information, the magnitude/implication/relevance of which, was large enough that it seemed like it was a substantial contributor or argument towards taking the action you selected. All correct and justified actions are reasonable, but not all reasonable decisions are justified or correct.
For example:
- It turned out you didn’t find any contraband, therefore it was not omniscient optimal to search the premises
- Given the urgency of the effort against organized crime, according to the aggressive strategy favored by the value system of your law-and-order superiors, it was correct to search
- Under the “probable cause” standard, where you acted on the tip you got from your informant that had been right three times in a row, clearly your actions were justified as greater than 50% probability
- You got information from a usually reliable informant, therefore you were reasonable in taking action based on that information