Managing Probabilistic Outcomes
In deterministic processes, outcomes are defined by the action of the agent.
Say you are playing tic-tac-toe or checkers. Even before you make your move, you can forecast ahead of time what this move means for the board state and what options are then made available for your opponent.
That's why these can be considered “solved” games. A computer, given any game state, can find the optimal play for the current game and force, move-by-move, the best available outcome.
In fact, by looking ahead in the so-called “search tree”, and assuming the opponent will make the best available move available to them, the best available outcome of the game can be predicted with total certainty.
Most of the processes in our lives, on the other hand, are not deterministic, but probabilistic (to be pedantic, they are stochastic).
When the agent takes an action, there's a distribution of possible outcomes.
For example, there may be a low probability things go bad and a medium/expected chance things go well, followed by a low probability of things going very well.
Even without mathematical training, we intuitively grasp that as risk and uncertainty.
Say, in one scenario, you are driving on a well maintained an quiet road. It's likely that you will feel relaxed and may let your attention drift a bit. On a second scenario, though, if there's intense traffic and the conditions are challenging, you will be way more attentive and careful with your driving.
The second scenario inherently feels more risky and unpredictable, so we respond to it by redoubling our attention and care.
Many of the decisions we have to make in life, likewise, are based on incomplete (and in fact unknowable) information, leading to uncertain outcomes.
What graduation path do I choose? Your employability will depend on the job market 4 or 5 years from now (an unknowable piece of information -- one can only hope currently observable trends are maintained).
Do I keep working on my current job or do I switch roles? The outcome of keeping the current role is more predictable (lower risk/variance in outcome), but maybe there's substantial upside in looking elsewhere (at a cost of higher risk and outcome variance).
Unlike in checkers and tic-tac-toe, deterministic games, there's no correct answer in a stochastic world.
Someone can make a good decision and still have things go badly. Someone can make a bad decision and have things go well.
Sound decision-making means making the best possible decision by collecting the necessary information and weighing in likely risks and possible outcomes with the currently available knowledge.
I put emphasis on currently available knowledge to drive in the point that the outcome may not be known ahead of time. Still, decisions have to be made, since not materialising any decision and doing nothing is an action in itself.
In the end of the day, though, you may only judge the soundness of your past decision by the available information at the time you took it, not by how well things in fact turned out, and the latter are for the most part unknown.