Uncertainty Is Not a Technical Problem

On uncertainty, risk modeling, and the incentives that reward precision over responsibility

We expect risk models to tell us what will happen.
When they don’t, we assume something went wrong.

But many of the failures we attribute to bad modeling have nothing to do with technical mistakes. They come from a deeper misunderstanding of uncertainty itself.

Some uncertainty exists because we lack information. Better data and better models can reduce it.
Other uncertainty exists because the world is inherently variable ~ storms, systems, and people do not repeat themselves. No amount of precision can eliminate that.

Modern institutions are not confused about this distinction. They are structured to suppress it. Precision offers clarity, defensibility, and comfort ~ even when it misrepresents reality.

This essay argues that uncertainty is not a technical problem.

It is a moral and institutional one.

Uncertainty Is Not the Same as Ignorance

We tend to treat uncertainty as a failure of knowledge. If only we had better data, better models, better tools ~ then the answer would reveal itself.

Sometimes that’s true. Some uncertainty reflects limits in what we know: missing data, outdated assumptions, imperfect measurements. In principle, that kind of uncertainty can be reduced.

But other uncertainty reflects real variability in the world itself. Storms do not repeat themselves. Systems respond differently depending on timing and context. People behave differently even when faced with similar conditions. Even when models perform exactly as designed, outcomes can diverge widely because the sequence and interaction of events ~ not just their magnitude ~ shapes consequences.

No amount of information eliminates that variability. At best, it can describe its range.

The mistake is assuming that all uncertainty belongs to the first category ~ that every unknown is simply waiting to be resolved by better measurement. When reality proves otherwise, frustration sets in.

Someone must have done something wrong.

Why We Confuse Precision With Truth

A precise number feels accountable. It looks like work has been done. It suggests that judgment has already been exercised by someone else, somewhere upstream.

A range, on the other hand, demands participation. It forces us to ask uncomfortable questions: Which outcome matters most? How bad is bad enough? What are we willing to tolerate?

Those are not technical questions.

They are moral ones.

And modern systems are not designed to answer them well. So instead, we ask numbers to do the work of judgment for us.

Why Institutions Reward False Precision

This drift is not usually malicious. It is structural.

Institutions reward outputs that travel cleanly ~ numbers that fit neatly into briefings, tables, and approval memos. They penalize caveats, ranges, and statements that admit limits. A clean estimate is easier to defend than an honest one.

Precision simplifies responsibility. If a number is “the estimate,” then accountability can be traced to the method rather than the decision. When outcomes diverge from expectations, the failure can be framed as technical rather than judgmental.

This is not because analysts do not understand uncertainty. Most do. It is because expressing uncertainty honestly often creates friction the system is not built to absorb.

So uncertainty is compressed.

Not eliminated ~ just hidden.

How Precision Shifts Responsibility

Precision does more than describe reality. It distributes blame.

A single number allows responsibility to be outsourced to process. It implies that someone else ~ an expert, a model, a method ~ has already done the hard thinking. The moral weight shifts from decision-makers to calculations.

This is why precision is comforting even when it misrepresents reality. It creates the appearance of control. It signals seriousness.

It offers moral cover.

But when reality diverges ~ and it always does ~

that cover dissolves.

Why Better Data Doesn’t Always Improve Decisions

Modern systems are very good at reducing certain kinds of uncertainty. They can refine measurements, update inventories, recalibrate models, and improve resolution. Each improvement produces a more confident-looking answer.

But not all uncertainty matters equally.

Some uncertainty dominates outcomes regardless of how refined the inputs become. Variability overwhelms marginal precision. Rare events shape consequences far more than averages. Human behavior refuses to conform to tidy assumptions.

In decision analysis, this is the problem of value of information. Reducing uncertainty has a cost, and the benefits of additional refinement diminish once it no longer changes choices.

Beyond that point, further precision improves confidence more than it improves decisions. Resources are spent perfecting what can be perfected rather than confronting what cannot be controlled.

This is not a technical failure.

It is an incentive problem.

When Risk Estimates Erode Public Trust

Public trust erodes not ‘cause estimates are wrong, but because expectations are misaligned.

When a precise number is presented, people assume it carries a promise: This is what will happen. When reality deviates, the assumption becomes accusation. Either the experts were incompetent, or they were dishonest.

Institutions respond defensively, pointing to methods, margins, and footnotes that were never part of the public understanding in the first place.

Both sides talk past each other.

The numbers were never meant to predict a single outcome ~ but they were presented as if they did.

The failure was not analytical.

It was translational.

A Better Way to Think About Risk and Uncertainty

Instead of asking whether an estimate is right, a better question is ~

What are you not-so-sure about?

Is the uncertainty reducible with more information, or does it reflect real variability in the world? Does refining the estimate change decisions, or just make them feel safer?

What uncertainty has been suppressed to make the answer sound better?

What are you not telling me?

Analysis cannot eliminate judgment ~ but it can inform it ~

if we are honest about the uncertainty.

What Risk Models Are Asked to Hide

Uncertainty is not a flaw in the system. It is a feature of reality.

The problem is not that uncertainty exists, but that we keep asking technical tools to solve moral problems. We ask numbers to absorb responsibility, to stand in for wisdom.

Eventually, reality comes knocking.

When it does, failure is blamed on expertise itself, rather than on the incentives that shape how expertise is expressed.

Until we confront that, we will keep demanding certainty where it cannot exist ~ and losing trust when God laughs at our scribble, lines in the sand.

Author’s Note

This essay draws on concepts from decision and risk analysis, particularly the distinction between knowledge uncertainty (uncertainty arising from limits in data, models, and assumptions) and natural variability (uncertainty arising from real, irreducible variation in physical and human systems). It is informed by academic frameworks that treat uncertainty not merely as a technical issue, but as a central factor in decision-making, responsibility, and communication.

The aim here is not to reject quantitative analysis or modeling, but to clarify what such tools can ~ and cannot ~ do.

Precision has value.

So does honesty.

The tension between the two is not a flaw ~ it is the condition under which judgment must operate.