- LATEST BLOG POST
- echo $post_date ?>
- In a world that used to be data-starved, is data indigestion a risk? Data starvation It is rare that we…
- Read More
Today it’s kinda windy, some clouds, I feel just fine, but my wife thinks it’s cold when the sun goes incognito behind the clouds. Quite an uncertain weather, right?
Focus now on the possible downward outcome: rain later in the day, during our open air BBQ (rain is the hazard, BBQ is the part of the system which will be potentially hit).
We can measure the related uncertainty using the POP (Probability of Precipitation) released by the weather channel, or use some kind of calibrated judgment yielding a usable estimate: say 30% chances of rain. Of course it would be even better (do not dream about this) is we could have the probabilities linked to various magnitudes of the rainy event (1mm/hr, 10mm/hr, etc. for so many hours).
The consequences will be complex (like usual): people may be late disrupting the caterer organization and increasing our costs for catering and rented furniture; the caterer may be late, creating “image” problems; everyone may want to seek shelter in the house and we may incur carpet cleaning costs, and so on.
If we consider the average of those extra costs and sum them up after carefully checking they are not mutually exclusive or correlated to a certain level, then multiply by the probability, then we have developed the “expected opportunity loss” (EOL). Depending on the entity of the EOL we may decide it’s tolerable, or try either to mitigate (all or partial losses), or seek to transfer the risk by asking an insurance cover. An insurer would at least compute the EOL, add operating costs and profit and calculate the applicable premium.
What is left to do is discussing credible, possible, predictable and unpredictable. This is an important discussion to frame the lower bound of probabilities it is possible to include in a risk analysis and also the limit of what we call “an Act of God”, which is the key to activate Force Majeure clauses in many commercial contracts, including insurance ones.
We are confronted with limited knowledge on our present or future “environment”, where even the existing state is difficult to describe, let go future outcome(s).
We measure uncertainty by assigning probabilities (likelihoods) to the set of outcomes we imagine possible (credible outcome vs. possible outcome will be the subject of another discussion).
Due to a hazard (or possibly a benign event) hitting a system an outcome is caused. If we evaluate the possible losses (or benefits) for an outcome, and we couple them with the probability, then we have evaluated the downward (upward) risk of that outcome.
Then we compare that risk with a tolerance threshold so see if we should act or not upon it.
Hence we say we live in an “uncertain” world but we can definitely sleep well by making sure we get the best possible evaluations of probabilities and consequences.
Of course there are other definitions of uncertainty, probability and unpredictability, but, from a technical point of view, the ones above should work in most contexts.