The Scandal of Prediction: Why Forecasters Keep Their Jobs

Here's something that should bother you more than it probably does: professional forecasters are consistently wrong, and they keep their jobs anyway. Economists, political analysts, market strategists — the track record on prediction is, in the aggregate, embarrassing. And yet the prediction industry is thriving.

Nassim Taleb calls this the scandal of prediction. Not a scandal in the sense of a secret — it's completely public information that professional forecasters fail at their stated mission. The scandal is that nobody seems to care.

What the Scandal Actually Is

The scandal has two parts, and both are necessary to understand it.

First: most predictions from credentialed experts are not more accurate than chance — and some are systematically worse, because the expert's confident framing causes people to act on them more than they would act on acknowledged uncertainty. The harm isn't just that the predictions are wrong. It's that wrong predictions presented with confidence produce worse outcomes than honest acknowledgment of uncertainty would.

Second: the people making the wrong predictions face no meaningful consequences. The economist whose GDP forecast was 3% off and whose trade recommendations cost clients real money is asked back for next quarter's panel. The political analyst who was wrong about every major event in a decade still has a bestselling book and a cable news contract. The structure of the prediction industry insulates its practitioners from feedback.

Taleb's formulation: "Anyone voicing a forecast or expressing an opinion without something at risk has some element of phoniness. Unless he risks going down with the ship this would be like watching an adventure movie."

This is not a claim that all forecasters are dishonest. Many are smart, well-meaning people doing their best with limited information. The phoniness is structural, not personal — they're in a position where they don't share the downside of being wrong, and that structural insulation corrupts the signal.

Why Wrong Forecasters Keep Their Jobs

Understanding why failed forecasters aren't fired reveals something important about the incentive structure.

Prediction fills a psychological function that accuracy doesn't satisfy. When a major market event, political development, or crisis happens, people want an explanation and a forward-looking narrative. An expert who says "I genuinely don't know what will happen and neither does anyone else" satisfies nobody. An expert who says "I believe we're heading toward X because of A, B, and C" satisfies the demand for explanation even if X doesn't happen.

Consumers of expert predictions are not, primarily, trying to acquire actionable intelligence — even when they think they are. They're managing anxiety about an uncertain future. The prediction does that work regardless of its accuracy. Which means the expert who delivers confident predictions fulfills the job description better than the honest expert who delivers calibrated uncertainty.

"The ancients knew very well that the only way to understand events was to cause them."

This aphorism points at the deeper problem. Understanding complex events causally requires the ability to run experiments — to change variables and observe outcomes. We can't do that with most of the situations we most want predicted. We observe history once. We can't re-run the 2008 financial crisis with slightly different housing policy and see what happened. We can't observe the counterfactual. Which means the confident causal stories experts tell are usually post-hoc narratives, not tested hypotheses.

The Ancient Wisdom on Prediction

"For the ancients, forecasting historical events was an insult to the God(s); for me, it is an insult to man — that is, for some, to science."

Taleb is noting that pre-modern cultures had a better epistemological grasp on prediction than we do. The ancient prohibition on trying to predict the future wasn't superstition — it was a recognition that the future, especially at the level of historical events, is not the kind of thing human intelligence was designed to penetrate.

Modern culture took the opposite stance: we can predict if we have enough data, good models, and processing power. The last century provided an extended natural experiment on whether this is true. The results are not encouraging. The models got more complex. The data got larger. The predictions did not get better.

"They would take forecasting more seriously if it were pointed out to them that in Semitic languages the words for 'forecast' and 'prophecy' are the same."

The same epistemological status. We've built an industry around the idea that secular forecasting is different in kind from religious prophecy — that it has the structural reliability of science, not the structural unreliability of oracles. The evidence suggests otherwise.

What You Should Do Instead

The scandal of prediction doesn't mean you can't make decisions about the future. You can and must. It means you should structure your decisions to be robust to prediction failure — not dependent on a specific forecast being right.

"For Seneca, the Stoic sage should withdraw from public efforts when unheeded and the state is corrupt beyond repair. It is wiser to wait for self-destruction."

This is a decision strategy that doesn't require accurate prediction: position yourself to survive a wide range of outcomes rather than optimizing for the most likely scenario. The bet on the most-likely scenario is also a bet that the most-likely scenario will occur. In complex systems, it often doesn't.

Concretely: - Weight opinions from people who have positions in the outcome, not just analysis - Prefer decision strategies that remain good across multiple futures - When an expert tells you they know what's coming, discount the claim proportionally to how specific it is - Be especially skeptical of confidence that comes without acknowledgment of what would falsify the prediction

The forecaster who says "I'm 70% confident in X, and here's specifically what I'd need to see to lower that" is more trustworthy than the one who says "this is clearly what's going to happen." The first one is doing epistemics. The second one is doing theater.

For the full framework, read The Bed of Procrustes Explained.