A counter is a rate, not a number
In October the Missile Boat had a problem that looked like success. It did its job, which is sinking ships from a distance, and it did it against everything. At equal metal it beat Patrol Boats, beat the Aegis Frigate, beat a battleship, and beat a battleship with an escort. Only submarines stopped it, and only because it cannot fire under the water.
A unit that wins every fight on the surface is not a unit. It is the fleet. This is an account of how we took it apart, and what the exercise taught us about counters in general.
How we ask the question
We do not tune by feel, at least not first. Two squads start sixteen tiles apart on open deep water and attack-move into each other. We run each fight on five fixed seeds and record who wins and how much of the winner is left. A change is a row in a table before it is an opinion.
Before any change, six Missile Boats against thirteen Patrol Boats, the same metal on each side: the boats won five of five with 60 percent of their hit points left. Five boats against two Aegis: five of five, with 91 percent left. The frigates barely scratched them.
The tempting fixes
The natural move is to reach for the boat's own numbers. Make it cost more. Shorten its range. So we tried both.
At 350 metal instead of 320, the boats lost all five fights against Patrol Boats at equal cost. With a shorter range, the same. The Missile Boat has exactly one fight it is built to win, and it wins it narrowly; any trim that touched that margin did not weaken the boat, it deleted it.
This is worth saying plainly because it is the mistake we make most often. A unit that is too strong everywhere is usually not too strong. It is missing an enemy.
Giving it an enemy
So we gave the Aegis point defence: a launcher that shoots the boat's missiles out of the air, over itself and the hulls beside it.
The first version was a disaster in the other direction. Point defence always hits what it fires at. A launcher that replaces its rounds as fast as the boats fire them does not reduce the damage, it cancels it. Two Aegis took on five boats and finished untouched. We had not made a counter. We had made a wall.
What we were tuning, it turned out, was never a damage number. It was a leak: the boats' rate of fire, less the escort's rate of replacement. Above zero and missiles get through. Below it and none do. The whole design lives in how far above zero that difference sits.
The settings we shipped were a magazine of three interceptors, three seconds to replace each one, and a boat that reloads in five seconds where it used to take three and a half. Those numbers produce a sentence a player can use: a pack has to outnumber the magazine.
Three Missile Boats attack one Aegis and lose, though the frigate limps home with 9 percent of its hull. Four boats attack the same frigate and win with three quarters of their strength left. Somewhere between three and four is a decision, and it belongs to the player.
The triangle closes itself
The part we did not plan is the part I like most. Eleven Patrol Boats against two Aegis, equal metal: the Patrol Boats win all five. Point defence does nothing against a gun.
So the Missile Boat still sinks Patrol Boats, as it always did. The Aegis now stops Missile Boats. And Patrol Boats outgun the Aegis. Nobody wrote that third line into the game. It fell out of fixing the first two honestly, without a special case anywhere.
When a counter web is working, most of it is like this. You add one true thing, and two other relationships you already had start to mean something.
What we took from it
A few habits came out of that pass, and we have kept them.
- Measure the fight the unit exists for first. Every change to the Missile Boat was checked against Patrol Boats before anything else. If a fix breaks the one thing a unit is for, it is not a fix.
- Prefer a new answer to a smaller number. Nerfs make a roster flatter. An enemy makes it deeper.
- Look for the rate. Most counters that feel binary, all or nothing, are two rates sitting on the wrong side of each other. Find the difference and tune that.
- Write down what you tried. The table of settings that did not ship is longer than the one that did, and it has saved us from trying them again.
The Missile Boat changed by one number. Everything else that changed was around it.