Discontinuous Utility and the Jutland Paradox

On October 21st, 1904, the Russian Baltic Fleet fired on British fishing trawlers in the North Sea. They had mistaken the boats for Japanese torpedo craft—an absurd error, given that they were off the coast of Hull and the Japanese navy was an ocean away. One fisherman died. Britain nearly went to war with Russia over it.

The incident was a preview. That fleet, having sailed halfway around the world to fight Japan, would be annihilated at Tsushima the following spring. It was the largest naval battle since Trafalgar, and it ended in hours. The Japanese lost three torpedo boats. The Russians lost their entire Pacific strategy, their reputation as a European power, and the war.

Every admiralty in the world took notes. The British took more notes than most.

The problem with naval power before 1906 was that it was continuous.

You could rank every ship in the world on a spectrum. At the top sat the newest pre-dreadnought battleships—four heavy guns, perhaps a dozen secondaries, thick armor, moderate speed. Below them, older battleships with fewer guns or thinner belts. Below those, armored cruisers. Then protected cruisers. Then everything else.

The gradient was smooth. A newer ship was better than an older ship, but not categorically different. A fleet with ten battleships could lose to a fleet with twelve. Naval strategy was arithmetic.

Admiral John “Jackie” Fisher had a different theory.

Fisher had watched Tsushima closely. The battle confirmed what he already suspected: long-range gunnery was the future. The Japanese had opened fire at ranges of eight thousand yards or more. At that distance, only the largest guns could hit anything. The intermediate guns that pre-dreadnoughts bristled with—the six-inchers, the nine-pounders—were useless. They complicated targeting, added weight, and accomplished nothing.

Fisher’s conclusion: stop building compromise ships. Build a ship with nothing but the biggest guns. More speed. More armor where it mattered. Accept no trade-offs.

HMS Dreadnought launched in 1906. She carried ten 12-inch guns where her predecessors carried four. Her steam turbines made her faster than anything else afloat. She was built in a year, a feat of industrial coordination that announced British resolve as loudly as the ship itself.

And the moment she touched water, the gradient collapsed.

This is the part that interests me.

Dreadnought didn’t make other ships worse. The pre-dreadnought battleships in every navy still had the same guns, the same armor, the same range. They were physically identical to what they had been the day before.

But something had changed. In any battle involving a dreadnought, those older ships would be sunk before they could close to fighting range. The twelve-gun broadside would reduce them to scrap metal while they were still firing at water.

Naval strategists had a term for what happened: they called the older ships “pre-dreadnoughts.” The word itself encoded obsolescence. They weren’t “slightly worse battleships” or “second-tier capital ships.” They were artifacts of a previous era, useful only for coastal defense and colonial gunboat work. The Royal Navy had just spent decades building the world’s largest fleet, and most of it had been rendered worthless in an afternoon.

The math changed. Germany now needed to count only dreadnought-type ships when calculating whether they could challenge British sea power. Britain’s overwhelming numerical superiority vanished. Everyone started from zero.

This is what I think of as discontinuous utility: the moment when a small improvement in absolute capability creates a categorical break in relative value.

Here’s the twist: the dreadnoughts never fought.

On May 31st, 1916, the British Grand Fleet and the German High Seas Fleet met at Jutland. It was the largest naval battle in history. Both sides had spent a decade and billions of marks and pounds building dreadnought fleets for exactly this moment.

The battle lasted twelve hours. The British lost more ships. The Germans claimed victory. And then both fleets sailed home and never fought again.

The German High Seas Fleet spent the rest of the war in port. The British maintained their blockade. The dreadnoughts—those world-historical instruments of discontinuous advantage—sat at anchor, too valuable to risk. The actual naval war was fought by submarines, destroyers, and mines. The battleships that had restructured global strategy became, in practice, a fleet in being: useful only as a threat, never as a weapon.

This is the second lesson of discontinuous utility, and it’s harder to accept than the first.

The first lesson is that the gradient is an illusion—that second-best options become worthless when the frame shifts. That’s the exciting part. It justifies bold moves, concentrated bets, the willingness to abandon sunk costs.

The second lesson is that even after you’ve built the Dreadnought, you might not get to use it. The discontinuity changes everything except the outcome. Britain and Germany both understood that dreadnoughts had obsoleted all previous naval power. They both built massive fleets. And then they discovered that the new weapons were too valuable to deploy. The frame had shifted, but the new frame was stalemate.

Jutland wasn’t a failure of the dreadnought concept. The ships performed as designed. The discontinuity was real. But discontinuous advantage doesn’t guarantee discontinuous results. Sometimes you build the thing that changes everything, and everything stays the same.

The same logic operates at smaller scales.

Consider recruiting. Standard career advice treats job offers as roughly substitutable goods. Apply broadly, maximize the number of offers, then choose the best one. The implicit model is continuous: more offers means more options, and more options is always better.

But offers aren’t substitutable when they exist in different competitive frames.

A mediocre position at a mediocre company isn’t 0.7x as valuable as a strong position at a strong company. It’s playing a different game entirely. The mediocre position gives you mediocre colleagues, mediocre problems, and a mediocre trajectory. You learn to operate in that context. Your network becomes people who also took mediocre positions. Your resume becomes a story about that world.

Five mediocre offers don’t compound into one good one. They’re pre-dreadnoughts. They exist on a gradient that has become irrelevant.

The people who understand this treat recruiting differently. They don’t optimize for offer count. They identify the positions that would represent discontinuous jumps—the teams doing exceptional work, the roles with exceptional scope, the companies where being there at all changes what’s possible—and they focus everything on those. One offer at the right place is worth more than a dozen offers at places that don’t matter.

The risk, of course, is that you end up with nothing. The person who applies only to exceptional positions might strike out entirely. The person who applies broadly will probably land somewhere.

But “landing somewhere” might be the trap. The mediocre position is easy to accept because it looks like progress. You’re employed. You’re learning something. You’re building experience. All true, and all irrelevant if the experience you’re building doesn’t transfer to the game you actually want to play.

Standard economic theory models goods as substitutable along smooth curves. If I have a shirt that provides utility U=1.0 and a shirt that provides U=0.8, I should wear the inferior shirt some nonzero percentage of the time. Diminishing marginal utility ensures variety.

This framework assumes the goods are competing in the same game.

But Dreadnought didn’t compete with pre-dreadnoughts. She made them irrelevant to the competition that mattered. The pre-dreadnoughts still existed. They still had utility—for certain purposes, in certain contexts. But those purposes weren’t the ones anyone cared about.

Utility functions are local. They describe preferences within a competitive frame. When the frame shifts, the function doesn’t smoothly adjust. It breaks.

What does this mean in practice?

First: incumbency is worth less than it looks. The Royal Navy’s fleet of pre-dreadnoughts represented decades of investment and unmatched expertise. None of it translated to the new era. Germany, starting from scratch, could build dreadnoughts just as well as Britain could. The historical advantages evaporated.

Second: the transition is faster than you expect. Dreadnought launched in 1906. By 1914, every major navy had dreadnought fleets, and pre-dreadnoughts were afterthoughts. Eight years to completely restructure global naval power. Career transitions can be faster still—a single year at the right company can obsolete a decade of experience at the wrong ones.

Third: recognition lags reality. The German admiralty understood Dreadnought’s implications almost immediately—they had no choice, since their strategy depended on the calculation. But the people who had built pre-dreadnoughts, who had trained to fight in them, who had staked their careers on the old paradigm—they were slower to adjust. Some never did.

The same is true of careers. The person with five mediocre offers sees abundance. The person who understands discontinuous utility sees five positions on a gradient that doesn’t matter.

Fisher understood something important: in a discontinuous world, the second-best option isn’t an alternative. It’s a trap.

But Jutland understood something too. You can build the Dreadnought and still end up in stalemate. The discontinuity is real, but it doesn’t guarantee victory. It guarantees that the old game is over. What replaces it might be better, or worse, or just different in ways you didn’t anticipate.

Build the new thing anyway. Accept that everything you’ve accumulated might be worthless tomorrow. Accept, too, that the new thing might sit at anchor, too valuable to use, while the actual war is fought with submarines and mines.

The gradient is an illusion. The cliff is real. What’s on the other side—that, you only find out after you jump.