The Cannae Problem

It's August 2, 216 BCE. The Roman army stands in formation under the blistering Italian sun. Eight legions strong—the largest force Rome has ever fielded—nearly 80,000 men await the order to advance. Across the plain at Cannae stands Hannibal's army, outnumbered almost two-to-one. The Roman commanders, Consuls Varro and Paullus, feel confident. How could they not? They've assembled the greatest concentration of Roman military might in history specifically to crush this Carthaginian invader who has plagued Italy for two years.

By sunset, 50,000-70,000 Romans lie dead or captured. The remaining survivors flee in disarray. It is the worst defeat Rome will ever suffer.

What happened?

The Romans fell victim to "The Cannae Problem"—when an organization's conventional wisdom, its tried-and-true methodologies, its fundamental understanding of how the world works—becomes the very thing that destroys it. The Romans knew how to win battles.

They had a system.

It had worked for centuries.

And Hannibal used that system against them to devastating effect.

The Cannae Problem is not exclusive to ancient warfare. It stretches across institutions throughout history and into our modern world. When organizations, systems, movements and humans become victims of their own success, when they mistake "what worked before" for "what will always work," they create the conditions for catastrophic failure.

And often, like the Romans at Cannae, they can't even see it coming until they're completely surrounded.

The Perfect Roman Machine

The Roman military system was a wonder of the ancient world—an almost algorithmic approach to warfare that turned citizen-soldiers into an unstoppable force that had conquered most of Italy and defeated Carthage in the First Punic War.

The Roman legionary system used standardized equipment, standardized training, and standardized tactics. Roman soldiers fought in a checkerboard formation, with the front line engaging the enemy, then cycling to the back to rest while the next line moved forward. This rotation system allowed Romans to maintain constant pressure while preventing fatigue. Their formations were flexible, disciplined, and nearly impossible to break through directly.

For generations, this system worked brilliantly against every opponent. Tribal warriors from Gaul? Crushed. Greek phalanxes? Defeated. The mighty armies of Carthage? Sent packing in the First Punic War. Rome's military success reinforced their belief in their system. Victory after victory proved they were right.

Why change what works?

This is the first trap of the Cannae Problem.

Hannibal's Mental Model Hack

Hannibal Barca was one of the greatest military minds in history. He had spent years studying the Romans. He understood both their tactics and their psychology. He knew their strengths, but more importantly, he recognized the blind spots that came with their confidence.

At Cannae, Hannibal didn't try to beat the Romans at their own game. He completely reframed the problem. While the Romans arranged their troops in their typical deep formation (even deeper than usual that day), Hannibal deployed his army in an unusual concave formation, with his weakest troops in the center and his strongest on the flanks.

As the battle began, his center gradually gave ground—not in disarray, but in controlled retreat. The Romans, sensing victory, pushed forward harder, their formation bulging into Hannibal's retreating center. Roman commanders interpreted this as success: the enemy was faltering! Their system was working!

Except it wasn't. Hannibal's retreating center was deliberate. As the Romans pushed deeper, Hannibal's stronger forces on the flanks held firm, then gradually enveloped the Roman formation. The Roman army found itself surrounded, packed so densely that many couldn't even raise their weapons. What followed was slaughter on an industrial scale.

Hannibal had hacked the Roman mental model. He turned their confidence into overconfidence, their strength into weakness, their unstoppable forward momentum into a deadly trap. The Romans weren't prepared for this because such a defeat didn't fit within their understanding of how battles worked. Their mental model couldn't account for it.

The Cognitive Biases Behind Disaster

The Romans at Cannae suffered from several cognitive biases that we now recognize—and that still plague organizations today.

First, they exhibited classic confirmation bias. For centuries, the evidence had confirmed that their approach to warfare was superior. Every victory reinforced this belief. Contradictory evidence—like Hannibal's earlier victories at Trebia and Lake Trasimene—was rationalized away as anomalies caused by bad terrain or poor leadership, not fundamental flaws in the Roman system.

Second, they suffered from the curse of expertise. The Romans were expert practitioners of their style of warfare, but this expertise created blindness to alternative approaches. Their mental maps of how battles should unfold were so ingrained that they couldn't recognize when the territory had changed.

Third, they displayed normalization of deviance. As Hannibal's center retreated in an unusual way, the Romans normalized this unexplained phenomenon to fit their existing mental model: "The enemy is breaking, just as expected." They couldn't see the pattern shift because they were too committed to their understanding of how battles develop.

Finally, they exhibited profound groupthink. Roman military doctrine was so established that questioning it would mark a commander as ignorant or cowardly. Historical records suggest that Consul Paullus had reservations about the battle plan, but social pressure to conform to established doctrine—especially with his co-commander Varro's enthusiasm—prevented those concerns from altering their approach.

When your success becomes the seed of your destruction, that's the Cannae Problem in action.

Modern Cannae Moments

What does Cannae have to do with modern institutions?

Everything.

The pattern repeats:

Kodak and digital photography: Kodak actually invented the first digital camera in 1975. But the company was so committed to its film business model that it couldn't adapt its thinking when digital technology began to take over. Their mental model—"We sell film"—prevented them from seeing that they were in the memory business, not the chemical business. By the time they realized their error, they were surrounded by digital competitors.

Blockbuster vs. Netflix: Blockbuster's leadership couldn't break free from their retail store mental model. They saw their late fees as a vital revenue stream rather than a customer pain point. Netflix exploited this blind spot. Blockbuster had opportunities to buy Netflix for $50 million in 2000—pocket change for what was then a $3 billion company. But their conventional wisdom about how the video rental business worked prevented them from seeing Netflix as a serious threat until it was too late.

Nokia's smartphone collapse: Nokia dominated the cell phone market in the early 2000s, with a peak market share of around 50%. Their mental model was built on hardware expertise and incremental innovation. When Apple introduced the iPhone in 2007, Nokia's leadership initially dismissed it. Their conventional wisdom—that phones needed physical keyboards, that battery life was paramount, that their Symbian operating system was superior—left them vulnerable to Apple's completely different mental model centered on software experience. By the time Nokia recognized the paradigm shift, they were surrounded.

What makes these examples Cannae moments is not simply that these organizations made mistakes. Or even that those mistakes led to catastrophic failures.

It's that their past successes created the conditions for these mistakes.

Their expertise became their vulnerability.

Their conventional wisdom became the instrument of their downfall.

Exploiting the Gap: How Disruptors Win

Disruptors and innovators intuitively understand the Cannae Problem. They specifically look for gaps between established organizations' mental models and reality. These gaps represent enormous opportunities.

Hannibal didn't get lucky at Cannae. He deliberately engineered a situation that would exploit Roman orthodoxy. Disruptors don't succeed by playing the same game better—they change the game entirely.

IBM dominated computing in the 1970s with their mainframe business model—massive machines for massive institutions at massive prices. Their mental model was built around computing as a centralized, expensive corporate resource. When minicomputers emerged from companies like Digital Equipment Corporation (DEC), IBM initially dismissed them. Why would anyone want less powerful computers? But DEC wasn't competing with mainframes directly—they were creating an entirely new category for departments and smaller organizations that couldn't afford mainframes. By 1982, DEC had become the second-largest computer company in the world while IBM scrambled to enter the minicomputer market they had initially dismissed.

Then DEC itself fell victim to the same pattern. Their mental model became fixed on minicomputers as the perfect balance of power and usability. When personal computers emerged, DEC's CEO Ken Olsen famously dismissed them, reportedly saying, "There is no reason for any individual to have a computer in his home." DEC saw PCs as toys, not serious computing tools. By the time they recognized the threat, companies like Apple and Compaq had established insurmountable leads. DEC, once valued at $14 billion, was eventually sold to Compaq for $9 billion.

WordPerfect and Lotus 1-2-3 dominated word processing and spreadsheets in the DOS era through commanding feature sets and blazing performance. Their mental model was centered on standalone applications. When Windows emerged with its graphical interface, both companies initially dismissed it as slow and inefficient—they were optimizing for keystroke efficiency while Microsoft was changing the interface paradigm entirely. By the time they created viable Windows versions, Microsoft Office had already established dominance. WordPerfect Corporation, once valued at $1.4 billion, was sold to Novell for just $855 million in 1994, while Lotus was acquired by IBM.

Novell ruled network operating systems in the late 1980s and early 1990s with NetWare commanding over 70% market share. Their mental model centered on specialized network protocols and proprietary directory services. When TCP/IP and the internet emerged, Novell dismissed them as technically inferior to their IPX/SPX protocol. They were optimizing for local network performance while the world was shifting to a connected model where universal compatibility trumped raw speed. By the time Novell embraced internet protocols, Microsoft and others had undercut their position so severely that NetWare became irrelevant.

Encyclopedia Britannica, selling premium knowledge door-to-door for $1,500+ per set, was blindsided by Microsoft's Encarta CD-ROM encyclopedia launched in 1993 for a skimpy $99. Britannica's mental model equated quality and authority with physical heft and premium pricing. They couldn't imagine their meticulously crafted content being threatened by a CD-ROM with less content but superior searchability and multimedia. By 1996, Britannica's sales had collapsed from 117,000 to 55,000 sets annually, eventually forcing them to abandon their print edition entirely.

In each case, the disruptors identified the mental models constraining established players and deliberately exploited the blind spots those models created.

The established players - like the Romans at Cannae - couldn't even process what was happening to them before they were surrounded.

Breaking Free of the Cannae Trap

How can you avoid becoming the victim of your own conventional wisdom?

Some basic concepts:

1. Implement red teams and war-gaming: The U.S. military now regularly employs "red teams" specifically tasked with challenging existing assumptions and developing strategies that would defeat current approaches. This is an argument in favour of institutionalizing skepticism.
2. Study near-misses, not successes: We tend to study our successes and ignore close calls. But near-misses often contain valuable information about weaknesses in current approaches that were just lucky enough not to be catastrophic... yet.
3. Reward productive dissent: When everyone agrees, something is wrong. Actively reward people who challenge conventional wisdom productively.
4. Develop multiple mental models: Charlie Munger famously advocates for having a "latticework of mental models" rather than a single framework. Deliberately cultivate multiple ways of understanding your work and regularly test them against each other.
5. Practice temporal displacement: Ask, "If I were starting from scratch today, knowing what I know now, would I do things the same way?" If the answer is no, that's a red flag that conventional wisdom may be deviating from reality.

The Romans eventually adapted after Cannae.

They stopped trying to defeat Hannibal in pitched battles and adopted a strategy of delay and attrition under Fabius Maximus.

This "Fabian strategy" ultimately worked—not by doubling down on conventional wisdom but by recognizing that the mental model needed to change.

The Fundamental Attribution Error of Failure

When organizations experience their Cannae moment, the aftermath typically includes finger-pointing. At Cannae, Romans blamed Varro's leadership. At Kodak, people blamed CEO actions. At Nokia, observers cited executive decisions.

This is a fundamental attribution error. While leadership certainly matters, the Cannae Problem is primarily systemic, not individual. The issue isn't just bad decisions; it's that the entire decision-making framework—the mental model itself—has become obsolete while still appearing valid.

The Romans didn't lose at Cannae simply because Varro was a bad general (though he may have been). They lost because their entire understanding of how battles worked—an understanding that had served them well for centuries—suddenly failed catastrophically against an opponent who deliberately exploited its limitations.

Success creates its own failure mechanisms. The very things that make you successful produce the blind spots that make you vulnerable. Your greatest strengths, taken to their logical conclusion without adaptation, become your greatest weaknesses.

How many organizations today are standing in formation at their own Cannae, confident in their conventional wisdom, unaware that they're about to be surrounded? What mental models in your organization are so successful, so proven, so obviously correct that no one questions them anymore?

Those are exactly the ones you should examine first.

Because in the conceptual distance between your mental models and reality lies the space where your Hannibal is waiting.

What Would Fabius Do?

Rome's eventual strategy—the Fabian strategy of delay, harassment, and avoiding direct confrontation—wasn't intuitive to Romans. It felt wrong. It contradicted centuries of conventional wisdom about how Rome waged war. It required tremendous courage to implement because it appeared cowardly by traditional Roman standards.

But it worked.

The hardest mental models to change are those tied to identity. Rome's aggressive, direct military approach wasn't a strategy; it was part of the Roman identity. Similarly, Kodak's film business, Blockbuster's retail stores, and Nokia's hardware focus weren't business models—they were core to how these organizations understood themselves.

Breaking free of the Cannae Problem often requires being willing to feel like a traitor to your identity in the short term to ensure its survival in the long term. Fabius was initially mocked as "Fabius the Delayer," seen as cowardly for refusing direct battle. Only later was his approach recognized as the salvation of Rome.

The organizations that survive their Cannae moments are those willing to ask: "What would our Fabius do?" They're willing to consider approaches that feel wrong, that contradict conventional wisdom, that challenge organizational identity—because they recognize that sometimes survival requires abandoning the very things that made you successful in the first place.

The Cannae Problem will never go away. Success will always breed orthodoxy; orthodoxy will always create blind spots; someone will always exploit those blind spots. But awareness of this eternal cycle is the first step toward breaking it. The greatest tragedy of Cannae is not that it happened, but that versions of it keep happening, millennium after millennium, because we fail to learn its fundamental lesson: what worked before won't work forever, and success today creates the conditions for failure tomorrow.

Unless, of course, we're willing to surround our own conventional wisdom before someone else does it for us.

Postscript: Sometimes the Enemy Is Just Better

After the wreckage of Cannae, Rome blamed Varro. After Kodak collapsed, they blamed the CEO. After every systemic failure, we look for a scapegoat. We want a name. We want a culprit. And if we can’t find one, we invent one—because blaming a person is easier than admitting the other side was simply better.

Which brings me to Gettysburg.

In the aftermath of the Confederacy’s most devastating defeat, Southern newspapers and politicians scrambled to assign blame. General James Longstreet was cast as the Judas of the Lost Cause—accused of dragging his feet, disobeying orders, lacking “the spirit of attack.” Jubal Early pointed fingers. Others whispered about Lee’s declining health, or his divine fallibility. Pickett’s Charge, the suicidal climax of the battle, became a hot potato passed from one general to the next, as everyone tried to rewrite their own role into something nobler, less damning.

And then there was George Pickett himself, whose division was obliterated in the charge that bears his name. When asked why the South lost at Gettysburg, Pickett didn’t launch into a tirade about Longstreet or Lee or tactics or terrain.

He just said:

“I always thought the Yankees had something to do with it.”

It’s the most honest line anyone spoke about that battle. Because sometimes, yes—your system failed, your model was flawed, your leadership blinked. But sometimes?

The other guy just outmaneuvered you.

That’s the final blind spot in the Cannae Problem. The belief that all failure is self-inflicted. That with the right tweaks, the right model, the right war game, you could’ve won. It’s comforting. But it’s also a lie.

Sometimes you lose because someone like Hannibal is on the other side of the field. Someone who sees your system more clearly than you do. Someone who doesn’t just break your model—they turn it against you.

Pretending otherwise isn’t strategy. It’s denial. And it’s how you guarantee you’ll walk into the next trap with your eyes wide shut.