Lesson 5 of 16 · Chapter 5
Quantifying the Unexpected
Some places are simply harder to forecast than others. This lesson borrows two tools from information theory, entropy and Bayesian updating, to measure how open a future is and how far a new fact should move you, so you can plan honestly and change your mind in proportion.
Step one
Read the material
Five ideas. What entropy measures, and what it does not. The four layers inside the word surprise, and why plans fail in high-entropy places. Priors, evidence and posteriors. Updating in proportion instead of not at all or all at once. And the chaos margin that turns all of this into a better plan.
Idea one
Entropy: how open the future is
A published timetable is predictable enough to plan around. A reply after a job interview is far more open. A shuffled deck has clear outcomes, but you cannot know which card comes next. Information theory measures that structure with entropy. In everyday speech it means mess; here, entropy measures uncertainty in a set of possible outcomes.
A coin that always lands heads has almost none: seeing heads tells you nothing new. A fair coin has more, a die more still, a deck more again. The more possible outcomes there are, and the more evenly those outcomes are distributed, the greater the entropy. A 99 percent chance of clear skies holds little uncertainty; 40 percent sun, 30 rain, 20 snow and 10 wind holds much more. It is measured in bits: a fair coin flip is one bit, a die more than two.
Entropy does not tell us whether an event is good or bad. It tells us how much uncertainty existed before the event was known. A wonderful call can come from a low-entropy situation; bad news long expected can still bring grief.
Idea two
Four layers of surprise, and planning honestly
The word surprise bundles four things: the unexpectedness of one outcome; the overall uncertainty of the environment that produced it; its emotional meaning; and what must now be updated. Entropy is the second. A low-entropy commute runs on time with clear announcements; a high-entropy one changes without warning.
That changes how a wise person plans. Ten minutes of margin may do on a reliable line and be fragile on an unreliable one. This is not pessimism. It is accurate planning in the presence of uncertainty. Much human frustration comes from treating high-entropy environments as though they were low-entropy ones, then being shocked twice: by the disruption, and by discovering the plan quietly assumed a steadier world.
Uncertainty also makes us hungry for updates, which is wise when an announcement can help. But rereading a short message ten times cannot produce what has not arrived: no amount of repeated inspection can extract information that has not arrived. Ask what can be known now, what is truly unknown, what is likely versus merely possible. Sometimes the future is simply still carrying information we have not received.
Idea three
Priors, evidence, posteriors
Thomas Bayes gave us a principle for reasoning under uncertainty: begin with what you believe, receive evidence, and revise in proportion to how strongly it points elsewhere. The brain does this constantly, if roughly. Your starting beliefs are priors. A prior is not necessarily a prejudice or an error. On a line where delays are usually short, “this will probably resolve soon” is sensible.
Then evidence arrives: “service will be suspended for an indefinite period.” The revised belief is the posterior. Prior: a brief pause. Evidence: suspended indefinitely. Posterior: this is serious, and I need a new plan. Bayesian surprise is the size of the change between the prior and the posterior. A three-minute delay barely moves you; a line closed for months rewrites the map at several levels.
So surprise is not located entirely in the event. The same rainstorm surprises the person who trusted the forecast completely more than the one who saw the clouds and brought an umbrella. Wiser people are still surprised; they just do not treat 70 percent as a guarantee. The chapter’s word for it: distinguish confidence from certainty.
Idea four
Updating in proportion
A friend does not answer an invitation. The first thought: they no longer care. That conclusion is often too large for the evidence. There are two common errors. One is refusing to update: “Nothing has changed.” The other is updating too completely, too soon: “One unexpected event proves that everything I believed was false.” Both are rigid. It says that beliefs should move, but they should move in proportion to evidence.
The chapter’s three questions: What did I believe before this happened? What new evidence has actually arrived? How much does this evidence reasonably require me to revise? An employee given hard feedback need not conclude “my manager is wrong” or “I am terrible at my job.” Part of the work needs attention; the self-assessment becomes more accurate, not destroyed.
The hardest updates hit beliefs we did not know we held, like “this will not happen to me now.” Hoping was not foolish; life needs provisional trust. But a useful belief is not a fortress. It is a working estimate. This is Bayesian maturity: the ability to let evidence change you without allowing every new fact to erase you.
Idea five
The chaos margin and entropy budgeting
Data science cannot remove uncertainty; it teaches a disciplined relationship with it. The practical move is a chaos margin: space deliberately included in a plan for the uncertainty that reality is likely to produce. Extra time before the important meeting, savings for the unexpected repair, a backup of essential work, a second route. The margin does not mean that a person expects disaster.
This is entropy budgeting: allocating time, attention, energy and money for variation. Many people do it in one area and not another: an emergency fund, but a calendar so full one delayed train wrecks a week. Look for plans that quietly assume perfect conditions. A resilient plan says: this is what I will do if conditions hold, and this is what I will protect if they do not.
The same applies to feelings. The essential point is that a single data point is not always a complete story — though a pattern of evidence can justify a larger revision. At some point, wisdom means seeing the boundary between useful information and compulsive reassurance-seeking. Then ask: What is the next action supported by the information I have now? A well-made map does not promise that the territory will stay still. It helps the traveler recognize where they are when it moves.
Step two
Work it with your hands on it
Two working instruments. The Entropy Meter computes real entropy, in bits, for the chapter’s own examples, and for any four outcomes you set. The Update Bench takes the commuter’s prior, applies an announcement, and shows the posterior and the size of the Bayesian surprise.
Entropy Meter: choose a situation
Update Bench: the commuter’s prior
The evidence that arrives
The Entropy Meter uses Shannon’s formula exactly. On the Update Bench, the chance of each announcement under each explanation is an illustration chosen to match the chapter’s story (a three-minute notice fits a brief pause; an indefinite suspension fits serious trouble). Bayesian surprise is shown as the distance from prior to posterior in bits (relative entropy).
Hands on it, off the screen. Find one plan this week that quietly assumes perfect conditions: a calendar with no gap before the important thing, a budget with no room for a repair. Add one chaos margin to it.
Then, the next time something unexpected arrives, write the three lines: What did I believe before? What evidence has actually arrived? How much does it reasonably require me to revise — a small adjustment, a new plan, or more information first?
Step three
Prove you understood it
One question per idea, a new situation every time. Miss one and you are sent back to the idea it was testing.
All five solid.
You can say what entropy measures and what it does not, separate the four layers of a surprise, name a prior, the evidence and the posterior, update in proportion instead of never or all at once, and build a chaos margin into a plan. Chapter six turns to the illusion of complete data.
Surprise Is Constant
Lesson five of sixteen, from chapter five. Written by our founder, Dr. Gene A Constant, and donated to the Foundation. Published by Global Sovereign University Press.
Buy a copy — it funds a free seatGSU is a 501(c)(3) nonprofit and book sales are its only income. A purchase is not a tax-deductible donation. This lesson is free and always will be.