본문으로 건너뛰기
온라인 플레이

No hidden modifiers

·Rene Wang

Strategy games tend to accumulate secrets. A veteran bonus that is never printed. A flanking multiplier you learn from a forum thread. A hit chance that turns a won fight into a lost one and leaves you with nothing to learn from it.

Each of these was added for a good reason, usually depth. I think most of them subtract it. Depth is what a player can reason about. A rule you cannot see is not a rule you can plan around; it is weather.

The first pillar in our design document says: trust the player. Simple, legible rules that combine into deep play. No hidden modifiers. This post is about what that sentence has cost us, and why we keep paying.

One table

Every weapon in Steel Tide has a class, which is what it was built to kill. Every hull has an armour class. Damage is the weapon's number multiplied by one cell of one table, and that is the whole of it.

The table cuts both ways. An anti-tank missile does 0.4 of its damage to a light vehicle. A machine gun does 0.4 to heavy plate. A cannon is at 0.6 against soft targets. Nothing in the game is good at everything, and the reason is not a paragraph of exceptions, it is a grid you can hold in your head after a few matches.

The discipline is in not adding a second table. Whenever a fight comes out wrong, the cheapest fix is a special case: this unit takes less from that unit. We allow a single weapon to override a cell, and when it does, the resolved number is what the tooltip and the wiki print. A player should never have to choose between what the game tells them and what the game does.

No dice

Anti-air missiles in Steel Tide can miss. They do not miss by chance.

A guided missile flies 1.5 times its launcher's range, turns at a fixed rate, and loses its lock when the target passes more than 60 degrees off its nose. Once it loses the lock it flies straight until it expires. That is all. A fighter at cruise speed is caught by a missile fired from five tiles behind it. The same fighter on afterburner escapes the same shot. A crossing target can outrun the seeker's turn; a slow one, or one flying straight at the launcher, cannot.

We could have written this as a percentage. It would have been simpler to build and simpler to balance. But a percentage teaches nothing. When a pilot escapes a missile here, it is because of where they were and which way they turned, and when they do not, the replay shows them why. The outcome is an argument you can have with yourself.

The obvious move should be a right one

A new player facing a rush of light tanks does the obvious thing. They put a gun at the gate.

In a lot of strategy games this is quietly a mistake, and the game lets them find out later. We decided it would not be one. A gun emplacement in Steel Tide beats its own price in the units standing in front of it, on purpose, because it can neither move nor take ground. We keep a test that fails if that ever stops being true.

This is not the same as making defence strong. Artillery outranges every emplacement. A raid on the other approach ignores it. What it means is narrower: the first idea a reasonable person has should not be a trap. Mastery should come from learning more, not from unlearning what looked sensible.

Say what it is

The same rule reaches the words. A unit's description sits beside a panel that already prints its hit points, damage, range and cost, so the description carries none of them. It does not tell you how to use the unit. It does not rank it against the rest of the army.

It says what the thing is, and stops. The rest is the player's to find, and finding it is the game.

What it costs

Legibility is expensive in a specific way: it removes places to hide. A fuzzy system forgives a bad number, because nobody can prove it is bad. When every number is printed and every outcome repeats, a mistake in the table is a mistake everyone can see by the weekend.

We think that is the right trade. A game you can read is a game you can get better at, and it is also a game that can be told, precisely, where it is wrong.

Trust runs in both directions.