History creates the illusion of predictability once outcomes are already known. Only after events converge do patterns become obvious, narratives become clean, and parallels feel inevitable. Before that point, history is far less useful than most executives, investors, and commentators would like to believe.
In theory, history can help us avoid repeating known failures. But even then, it only works when we extract constrained rules grounded in structural invariants – the underlying forces that actually shape outcomes, like capital availability, incentive alignment, and physical constraints. Everything else is pattern matching dressed up as wisdom. And this is problematic, because we often use that dressed-up wisdom to both constrain our choices (“history says that’s a bad idea”) and to make foolish choices (“when in history has this failed?”).
Not Anti-History
I wrote this section after I wrote and edited everything else here. But wanted to insert it closer to the beginning.
Just for clarity, I’m not against learning history. To the contrary, it’s one of my favorite subjects. I took multiple semesters of history in college just for fun. I read books on history and love biographies. In no way would I discourage people from learning history, it is fascinating.
But, here is the thing. Overtime I’ve noticed a strong desire of people to use history to predict the future and it doesn’t seem to work. That hasn’t hurt my interest in history but it has opened my mind to the idea that perhaps history, as enjoyable as it is, isn’t really about defining the future. That maybe the whole “those who don’t learn history are doomed to repeat it” concept is wrong, that people are going to repeat history anyways – that their failure to take some set of historical facts into consideration is only obvious afterwards. My current model is that history is a really good way to learn about humans but it rarely translates forward as nicely as many people wish.
So back to the essay.
Why History Fails as a Predictive Tool
History runs backward, not forward.
We observe outcomes, then work our way back to explanations that feel coherent and satisfying. Once a story is complete, it becomes very difficult to remember how uncertain, chaotic, or ambiguous the situation actually was while it was unfolding. Events feel obvious in hindsight, even when they were not.
This is where naive pattern matching takes hold. We see something that looks familiar and assume the same outcome will follow. Similar rhetoric. Similar behavior. Similar fear or confidence. These surface similarities feel predictive because humans are good at recognizing patterns. That does not make them causal.
For decision-makers, this is dangerous. Pattern matching feels disciplined, but it is usually unfalsifiable. When a prediction fails, there is no clear way to invalidate the historical lesson that produced it. The explanation simply shifts. Different timing. Different actors. Same “lesson.” That is not analysis; it is narrative maintenance.
Structural Invariants vs. Surface Similarities
The most common historical errors come from confusing surface similarities with structural invariants.
Surface similarities are easy to spot: rising prices, political slogans, public fear or enthusiasm, media tone. These are soft signals. They are visually compelling but weakly predictive.
Structural invariants are harder to see but actually determine outcomes: capital availability and cost, incentive alignment, time delays and feedback loops, physical and regulatory constraints, human limits around coordination and trust. These are the forces that shape what happens.
History transfers forward through structure, not appearance. Yet we consistently reach for the visible patterns while ignoring the hidden machinery.
What History Can Do Within Limits
History has real value, but it is narrower than commonly assumed.
History can help us deduce rules about how systems behave under specific conditions. Those rules only transfer forward if their underlying constraints still hold. Without clearly defined limits, historical lessons cannot be tested, challenged, or responsibly applied.
If a historical insight cannot be falsified, it is not a rule; it is a story.
Extracting useful rules from history is difficult. It requires studying multiple instances, identifying what was structurally consistent across them, and explicitly defining the boundaries of applicability. Most historical analysis fails because it stops at recognition rather than moving on to constraint definition.
The harsh reality is that the conditions required for historical rules to apply are so narrow and hard to identify that in practice, most historical reasoning provides false confidence rather than genuine insight.
Two Examples of Unconstrained History
The Maginot Line: Strategy by Analogy
After World War I, France absorbed what seemed like a clear historical lesson: major war in Europe would again be slow, defensive, and attritional. Fortifications worked. Time favored the defender.
This was strategy by analogy – same adversary, same geography, same recent experience. Analogies feel smart because they’re grounded in experience. But they only work if you’ve defined their limits.
France’s failure was not ignorance of history but misuse of it. They never asked: “What conditions made defensive warfare dominant in WWI?” A proper structural analysis would have identified the constraints: limited mobility, slow coordination, secondary air power. It would have asked: “Do these conditions still hold?”
When mechanization, radio coordination, and combined-arms doctrine changed those constraints, the system exited the conditions under which the rule applied. The lesson didn’t fail because history was wrong. It failed because its limits were never articulated.
The Gold Standard: When Conditions Change
Throughout the 19th century, the gold standard appeared to stabilize economies. After repeated financial crises, a consensus emerged: monetary discipline through fixed exchange rates maximized growth through stability. Inflation was the primary threat. Credibility mattered above all else.
But no one clearly defined why the gold standard worked, so they couldn’t recognize when the rule would no longer apply. The gold standard succeeded under specific conditions: limited capital mobility, slower economic transmission, smaller financial systems relative to GDP.
By the early 20th century, those conditions had fundamentally changed. Capital moved faster. Leverage was higher. Economic shocks transmitted instantly across borders. Policy response speed mattered more than signaling discipline.
Maintaining gold convertibility under these new conditions amplified economic contraction rather than preventing it. Countries that abandoned the gold standard recovered faster from the Great Depression than those that clung to it. The historical rule hadn’t failed – the system had moved outside the boundaries where that rule applied.
A Diagnostic Framework for Using History Responsibly
If history is going to inform forward-looking decisions, it must be treated as a diagnostic tool, not a prophecy.
- What must be true for this historical outcome to occur?
Identify necessary conditions, not visible symptoms. - Which conditions are hard constraints versus soft pressures?
Separate what cannot be violated from what can adapt. - Across which dimensions does this rule apply?
Economic, political, social, technological, temporal. - Where are the explicit boundaries of applicability?
If you cannot say where the rule stops working, you do not understand it. - What observable signals would indicate a breakout?
Define how you would know the system has exited the valid range. - If the outcome differs, could this rule be invalidated?
If failure does not falsify it, the rule was never real.
Most historical lessons will not survive this process. That is the point. History is not science; it doesn’t lend itself to universal principles and laws. The rules we can extract from studying history are few and their application is limited by real constraints and boundaries.
History as Discipline, Not Crutch
New outcomes do not mean history was wrong. They mean conditions changed.
History constrains behavior, not possibility. New combinations of incentives, technology, policy, and human behavior can produce results with no clean historical analog. Treating those outcomes as violations of history misses the structural shift underneath them.
Over-reliance on history often blinds decision-makers to novelty. Familiarity feels safe. It rarely is.
In the future, nations will fall, economies will collapse, and industries will disappear – that is inevitable. When those things happen, smart people will write books finding comparisons to past events. But when they try to apply those comparisons to predict future situations without defining necessary conditions, hard constraints, or breakout ranges, their historical analysis becomes intelligent-sounding guessing.
The Real Lesson
History deserves less reverence and more rigor.
Used properly, it helps identify constraints, understand failure modes, and avoid repeating known mistakes under similar conditions. Used poorly, it becomes narrative comfort mistaken for foresight.
The goal is not to predict the future by analogy. It is to recognize when historical rules apply, when they no longer do, and when something genuinely new is unfolding. Understanding structural constraints won’t tell you what will happen, but it can reveal what’s possible and what’s not within a given system.
Perhaps the most valuable historical lesson is this: systems are more fragile and conditions more temporary than they appear. Plan for optionality not because the future is random, but because your ability to identify which historical constraints still apply is limited. Build resilience to multiple futures rather than betting on historical patterns repeating.
That may be unsatisfying if you want history to provide comfortable predictions. But recognizing the limits of historical reasoning – while still extracting what constrained insights we can – is far more valuable than false confidence in patterns that no longer apply.
