The Barbarian Invasion Behind Your Statistics

Intangibles don't appear in white papers.

408 A.D. — Alaric leads his Gothic horde into Italy, ravaging all the way.

2008 A.D. — An invisible tangle of bad investments and government policy wreaks economic havoc in the United States.

What do they have in common? Statistics suggesting they won’t amount to much.

I was thinking about this circa 2008 because an editor assigned me background research on the economic chaos unfolding around us.

I pored over several years of Federal Reserve minutes and statements from Treasury secretaries. Alan Greenspan, Ben Bernanke, and Henry Paulson had trotted out statistics by the dozen to prove that the crisis some detected in the tea leaves would be minor, if it happened at all. The American economy had “strong fundamentals” and would shrug off a downturn like a head cold. But the waves of layoffs and cratering stock market in September suggested the experts missed something.

I kept returning to how convincing their positive stats sounded and how irrelevant they turned out to be. If they weren’t relevant, what was?

In my free time I was reading Edward Gibbon’s The History of the Decline and Fall of the Roman Empire, and he set up a marvelous comparison that helped me make sense of it.

Gibbon gives us “stats” for Roman-controlled Italy in 218 B.C., right before Hannibal Barca led his army over the Alps, set on humbling the growing Republic. Then he gives “stats” for 600 years later in 408 A.D., right before Alaric and his Goths invaded.

Any way Gibbon slices it, the later Romans seemed best prepared. Larger population. Bigger economy. Better state capacity and tax-raising ability. An expansive and wealthy noble class who might pitch in. Better technology. Weapon factories capable of equipping entire armies. And the list goes on.

But the stats were wrong.

When Hannibal trapped and slaughtered 70,000 Roman soldiers at Cannae and 20% of the male population of Rome vanished in a day, the survivors kept raising armies as fast as Hannibal could mow them down. They outlasted him. They wore away at his finite supply of soldiers and punished Italians who dared to side with him. The Romans would not surrender, no matter how badly Hannibal mauled them, no matter if the war took their last denarius.

Their descendants — better equipped in every essential — weren’t able to raise a single army as Alaric fell on them. They found no worthy commanders to lead their scattered garrisons; no one wanted to stick their neck out. The bureaucracy united only to bring down the half-barbarian general who’d proven Alaric’s only able competitor. When the Goths came for Rome, the Eternal City surrendered almost without a fight. A faction even threw the gates open for the invaders.

And what of Italy in 2025? Would it provide the strongest opposition of all to a new invader? It’s richer, more populous, and has high-tech armed forces to protect it. All the stats suggest it would put up the strongest fight, but I think Gibbon would call bullshit on our confidence. A survey found that only 16% of Italian adults would fight for their country if it were attacked¹. It must have been the same in the 5th century. Why risk yourself when you can pay barbarian mercenaries to defend you? Does it even matter if a Roman or a Barbarian emperor is in charge? Ambivalence is the greatest solvent of advantage.

Gibbon’s point is that there are often no stats for the unquantifiable essential — in this case, an I’ll give my life and fortune for it commitment. Is the average person an integral part of their country, or only along for the ride as long as it doesn’t get too bumpy? Without knowing, our conclusions are built on confident-sounding guesswork that might lead to disaster.

If you transported one of those hoary survivors of Hannibal’s victories who kept signing on for more campaigns into 5th or 21st-century Italy, they would spot the problem immediately — something was missing from the people.

Social cohesion, noblesse oblige, patriotism, and societal buy-in are only vaguely measurable. So we’re presented with what’s easily quantified as if it’s the worthiest measure. We get proxies of dubious significance and are led astray.

Let’s look at this another way. Today, we assume meritocracy is always good and nepotism is always bad. Binaries. We like those. But how much of that is based on having limited data? During the Napoleonic Wars, the British sank French ships at rate of 7:1 on average, and 32:1 for battleships². The explanation we get is that British captains were selected on merit while the French ones were nepo-babies.

That’s only half true. Yes, would-be British officers had to survive six hard years of training aboard ship before passing a taxing exam at the Admiralty. That’s the merit part — the measurable part — to weed out the obvious slackers and incompetents. But how do you choose captains from among these competent officers? The admiralty’s method looked a lot like nepotism.

The man who sank the most Frenchmen in this era was Vice Admiral Horatio Nelson. He’d served under an earlier era of officers prizing order-following from top to bottom above all else. That played out disasterously, since before radios it was often impossible to receive and implement orders in the smoke and chaos of battle. Nelson was the product of a move away from this. The admirals of his era wanted captain-level decision-making and tactical aggression. They felt you couldn’t go wrong if you went for the enemy’s throat, and that aggressive captains would best cautious ones.

But how do you find these bold officers? Exams weed out incompetence, but can’t detect unfakeable aggression. The solution was having successful officers recognize what statistics couldn’t uncover. To do this, they were necessarily promoting from those they had close ties with, often family members. It takes one to know one, in other words. Talent can spot talent even if the stats are silent on the matter. This sounds like a recipe for choosing incompetent relatives, but the results suggest otherwise.

In a study of the Royal Navy between 1690 and 1849³, drawing on performance records from 5,800 naval officers, those with family ties promoted to captain outperformed unconnected ones on the most important metric — sunk ships. Better assignments, crews, or equipment were ruled out as factors, and there was heterogeneity among the admirals making selections.

The captains with personal connections to admirals were 23 percentage points more likely to capture or sink an enemy than unconnected officers in chance encounters on the high seas. They were also less likely to fight an indecisive battle. The authors note, “essentially, connected officers either lost or won and never experienced a draw; whereas nearly a fourth of all unconnected officers fought indecisive engagements.”

How many outcomes in life hinge on factors that can’t be quantified, and which we’re likely completely ignorant of?

There were “unknowns stats” behind the 2008 recession too. Regulators and officials knew there was a parallel shadow banking system rivaling the public one in size, but since they didn’t have a way to measure it, they paid it no attention. That made their declarations of “solid fundamentals” worthless.

Now, you want me to tell you how to fix this. I can’t.

But maybe we can draw some conclusions:

• Many talking heads will turn out to be wrong, but we won’t know which ones in advance. We also don’t have the time to gain sufficient expertise to correctly judge for ourselves in most areas. So it’s best to minimize contact with talking-head ideas to prevent a false sense that you know what’s going on. News intake can be minimized as we restrict our reading to fundamental principles. Read old books.

• We’re unlikely to find infallible truth, but we can get incrementally closer to mastery and expertise in a few chosen areas.

• Data is great. It’s a useful guide for decisions, so give it its due. But we’re ever blinded by the stats we have, which glare so brightly that we often struggle to imagine the body of the relevant data we don’t have.

• When data backs a particular ideology/dogma/identity, it ends up making us stupider by rendering the belief itself painful to critically examine.

You may find these conclusions disappointing. Maybe you want me to say, “just call on a “it takes one to know one,” level of expert. But unless we’re one of those people, we can’t spot them from among the crowd of gossipmongers and keyboard warriors.

So basically, I’m saying be humble and be skeptical. That’s a decision that rarely lets us down.

And let’s hope there aren’t any barbarian hordes waiting in the wings, because our stats might not be what we think they are.

---

Thanks for reading Socratic State of Mind.

If you liked this article, please like and share it, which helps more readers find my work.

1. Reuters (2025) Only 16% of Italians would fight for their country, survey shows, 18 July.

2. Allen, D. W. (2002): “The British Navy Rules: Monitoring and Incompatible Incentives in the Age of Fighting Sail,” Explorations in Economic History, 39, 204–231.

3. Voth, Joachim, and Guo Xu. “Patronage for Productivity: Selection and Performance in the Age of Sail.” January 17, 2020