Decisions are not outcomes

On Sun, 11 Dec 2022, by @lucasdicioccio, 1395 words, 4 code snippets, 2 links, 1images.

Decisions are commodities of our lives: from the second we wake up and until we fall asleep, we are bound to make decisions to progress during the day. Most decisions we make are inconsequential: which shirt to pick, what flavour of ice cream to eat, and so on and so forth. As a result, we may easily fool ourselves that we are good at making decisions. In practice, gauging whether a decision is good or not is more difficult than it appears. In particular, we tend to conflate good decisions with good outcomes.

Decisions and situations

If we want to improve our decision making, we need to train ourselves to assess the quality of decisions independently from the outcomes of the decisions.

When we think what making good decisions entails, we often think about decisions that will improve things. And a common mistake people do when gauging decisions is to gauge the outcome. Indeed, we also make decisions that leave us indifferent. And, alas, sometimes we need to make the least worst decision.

Let’s illustrate these three situations with a series of trivial examples. For each example, you have two options: A or B and only one of the option can be chosen.


- gain $50

- gain $100

- gain $100,000,050

- gain $100,000,100

- lose $100

- lose $50

In each of the three situations above, option-B is always more favorable than option-A. As a result, someone making good decisions would always pick option-B over option-A. Furthermore, option-B is always $50 better than option-A, somehow, to most people in Situation-2, either options will feel indifferent.

An important observation in the above example is that any outcome of Situation-2 is preferred over any outcome of Situation-1, which in turn is preferred over any outcome of Situation-3. As a result, a person in Situation-3 taking option-B makes better decisions than a person in Situation-2 taking option-A! meanwhile, the person in Situation-2 taking option-A is in a significantly better place than the person in Situation-3 taking option-B.

The three situations above teach us the following: how-good a decision is independent from how-good an outcome is. Further, to gauge whether a decision is good or not, one must characterize the set of available options and then compare options with each other. In short: given a same situation, would you pick the same choice or not? Comparing options differs from comparing unreachable situations with each other.

Distinguishing between decisions and situation-changes

The key to distinguish between attainable options and unreachable situations is to understand whether an outcome is too-good-to-be-true ™️. Rational actors would switch to a better situation if they could because chosing poorly in a good situation results in better outcomes than chosing perfectly in a poor situation.

Thus, a first characteristics is that situations define a range of outcomes whereas decisions merely move the needle withing a range of outcomes. Thus, two situation-changing “decisions” are easy to distinguish because there is some boundary (or minimal overlap) between the ranges of achievable outcomes after taking a decision.

For people with a more “visual mind”, we could depict these differences as follows. Three situations (the range [----]) with a needle (the -o-). Options would move the needle around whereas situation changes would flip your situation to another row.

                           range of possible outcomes
situation Alpha:    [-------------o--------]
situation Beta:                          [--------o-]
situation Gamma:                                 [o--------------]

The ranges in the example above characterize different situations: the available choices for routine decisions are close to each other and result in marginal changes of outcome. Thus, hypotetically, after taking a first option and re-assessing the situation: would your routine decisions (for which you are mostly indifferent) fall on a similar continuum as if you had taken another option first? If you answer yes, the option you are assessing is a situation-changing option.

Note, if you are a manager of technical people, suggesting that switching situations is a mere decition will cross your interlocutor because hypothesizing unrealistic options implies an obvious situation-changing choice is not being made (i.e., it implies that your interlocutor is stupid). Questions like “why don’t you just replace system-XYZ” will be received as a sort of “stop-being-poor” insult or as intellectual-laziness. Not be the brainstorm-session ice-breaker you hoped for.

Alas, lines are often blurry. In business settings especially, situations that are unreachable to one may be reachable at an upper management echelon. In particular, the role of strategy is to put you in good situations, whereas the role of tactics is to make good decisions within your current situation.

Some business-settings examples of decisions among available options could be:

  • hire a new employee or keep on course
  • implement a piece of logic inside the DB or in the controller or in the model or in the view
  • build anew or maintain a system
  • should I investigate an issue by looking at changes or looking at logs or doing some reproduction checks

Some situation-changing decisions between unreachable (to most except C-suite people) situations could be:

  • stop paying for mission-critical systems to build in house
  • replace the whole team or keep on course
  • stop or not a money-making business-line with low profit-margins

In my experience, people mixing available options and unreachable situations happens on a regular basis. Probably not in extreme cases like my example above, where people recognize that transitory situations will exist. Making realistic plans require hard work to draw a picture of how a company could move from one situation to another one.

Making better decisions

Now we understand that a situation constrains the set of choices one can make. Within this set of choice, then we still want to make good decisions.

To make good decisions you need to start with enumerating available options, and enumerating and evaluating options is almost all there is to good-decisions making. Alas the enumeration itself can take too long or the evaluation may fall short due to lack of information.

If we draw these two dimensions on a board, we can split decision-making in four broad quadrants.

decision quadrants along evaluation difficulty and enumeration difficulty axes

Bottom-left are “easy decisions”: in these cases, intuitive choices are good enough. You can easily convince yourself which option is better and people coming after you gauging and revisiting your choices with hindsight will easily realize you were right.

Top-left are high-cost decisions that are amenable to scrutinization. Many (software) engineering situations fall in this quadrant: choices are few but evaluation is difficult. In this quadrant, I recommend to use evaluation matrices and other tools like design docs (you’ll find a template on the talks-and-docs page) to formalize and record how decisions were made.

Bottom-right are low-cost but numerous decisions. These choice happen when a seemingly-single choice actually consists of a combination of choices and where evaluation can be automated without too much guesswork. optimizations frameworks allows to enumerate and compare choices efficiently. For these techniques to work on needs to restrict the decisions to some formal and limited model of our reality.

Finally, in the difficult top-right quadrant, you may not be able to make a good decisions. The only reassuring thing is that if you cannot devise a good choice, your competitors likely are in a similar situation. In such a situation what is key is to understand the range of outcomes in best/worst/average situations, and assessing whether investing efforts in better decisions is offset by the outcome. Thus, you need some preliminary investigation to justify whether rolling a dice is an okay-move or not. If high-stakes are at play (that is if the range of outcomes are such that making a good decision is actually critical), I’d say that the best thing to do is to seek help from people with modeling, simulation, and/or scientific-experiments experience.


Decisions are choices in a set of options. Circumstances from your situation determine the set of options available to you. Empathize with people (including your past self) to disambiguate situations having poor outcomes from poor decisions. What is key is: given the same input and the same situation, would I still chose the same option?

In business settings, be careful to distinguish when decisions are tactical (i.e., trying to improve the outcome by making some choices) from when decisions are strategical (i.e., trying to improve the outcome by changing the range of choices). When the strategy attempts such a transformation, often what is amiss is a picture to move smoothly between two disconnected situations.

When deciding within options, your decision-making quality is a direct output of how thorough is your assessment of the set of available options (the enumeration) and your assessment of how good are options (the evaluation). Once you have these, making a good decision is a matter of ranking among choices. Keep in mind that analyzing choices has a cost, and the effort made to enumerate and rank choices should be smaller than the difference in outcomes between good and bad decisions. To help during this process, you should rely on decision frameworks like design-docs and mathematical modeling. Experts can help you with these.

I hope that, in retrospect, you’ll think that reading this article was a good decision 😬. Have a good day!