Public Opinion

The Post-Covid-19 Surprising Contingencies

Can anyone truly proclaim accurately the exact strategies that will work post-covid-19? Humanly speaking, I think not! We can sure make calculated guesses and chart our entry/exit strategies but we must prepare to have surprising contingencies. Let me give this analogy:

In reinforcement learning, a subset of machine learning, a model learns by interacting with a dynamic environment where it is expected to take a sequence of actions with a goal to maximise cumulative reward.

For each action, it gets an immediate feedback (or reinforcement) from an “interpreter”: a positive reinforcement (reward) for a right action (e.g., correctly predicting something), and a negative reinforcement (punishment) for an incorrect one.

At the start, the model is fed no historical #data for it to learn from, thus, it has no way of unambiguously knowing beforehand which actions are optimal. Simply, it discovers knowledge of the right or wrong actions by “learning on the job”.

In the same vein, the post-covid-19 phase-space is a completely uncharted territory for everyone. Even a time traveller from 20th century with post-influenza success record will stop dead in tracks when s(he) gets to 2020.

Just like in reinforcement learning, organisations and individuals will need to take several stochastic decisions by trial and error, where the goal is to maximise positive reinforcements or rewards. Only afterwards can they begin to choose actions as a function of the new history.

The more the rewards, the cleverer their mechanisms of finding trade-offs between the exploration of uncharted territory and the exploitation of current knowledge.

Ultimately, the dividing asunder between the champions and the spectators will become sharply apparent, the differentiators mostly being high-level insight, doggedness and risk tolerance.

Zion Pibowei

Leave a Reply

Your email address will not be published. Required fields are marked *

Follow by Email