Dune II: The RTS Template

There is a short list of games that invented something and then had the bad luck to do it so completely that everyone forgot it was an invention. Dune II is at the top of it. Base building, a resource you send harvesters out to collect, a tech tree that gates the good units behind the boring buildings, black fog over the ground you have not walked on, a mouse cursor that selects a unit and a right-click that tells it where to go, a sidebar down one edge with the build queue in it. Every one of those was decided by Westwood Studios in 1992, and thirty-four years later they are the water an entire genre swims in.

The full title was Dune II: The Building of a Dynasty, published by Virgin, and the “II” is doing heavy lifting — the first Dune game, from Cryo, was a lovely adventure-strategy hybrid with almost nothing in common with this. Westwood’s version came to DOS at the end of 1992, and the Amiga port followed. I met it on the Amiga, which was the wrong way round in every respect, and it still rearranged how I thought about strategy games in an afternoon.

What was actually new

The received wisdom is that Dune II invented real-time strategy, and that is too generous. Herzog Zwei on the Mega Drive got there in 1989 with real-time unit production and a contested map, and there are wargames earlier still that ran on a clock. Populous in 1989 had already established that you could point at terrain in real time and have a strategy layer respond.

What Dune II invented is the interface and economy that the genre actually uses. That is a smaller claim and a much more consequential one.

Consider the loop it assembled. Spice lies on the sand in visible patches. You build a refinery, which comes with a harvester, which drives out to the spice, fills up, and drives back. Credits appear. Credits buy buildings. Buildings unlock units and better buildings. Some buildings need to be adjacent to concrete you have already laid, or they degrade. Windtraps feed the whole base power, and a shortfall throttles everything you own. Meanwhile the map is dark until you push units into it, and the enemy is running the same economy on the other side of the dark.

Every single one of those pieces creates a decision with a cost. The harvester is the cleverest: it makes your income a physical object on the map that can be shot, and it means the economy has a geography. Your income is a convoy route. Defend the route or lose the war. Nobody had put an economic engine on the battlefield as a killable truck before, and it turns out to be the reason RTS maps have a shape at all.

The tech tree does the pacing work. It puts a decision in front of you at minute two — spend now on units that let you survive minute five, or spend on the structure that unlocks the units for minute fifteen — and that tension between economy and army is still the fundamental RTS question in 2026. StarCraft, Age of Empires, Command & Conquer, Company of Heroes, every one of them is a variation on the tempo problem Dune II posed.

And the mouse. This is easy to skim past because it seems obvious now. The strategy games before it were keyboard-driven, hex-based, menu-heavy things where you selected a unit from a list and issued an order from another list. Westwood let you point at a thing on the map and then point at where you wanted it. That is a small change that turns an accounting exercise into a physical act, and it is the reason the genre could ever be played fast.

Three houses that actually differed

The three-house structure — Atreides, Harkonnen, and the Ordos, invented for the game out of the deep apocrypha of the Dune Encyclopedia — earns its place because the asymmetry is real.

House Atreides get the Sonic Tank and the Fremen, and they play as the straightforward one with tricks. Harkonnen get the Devastator and the Deathhand missile, and they play like a brick to the head: heavier, slower, no finesse required. Ordos get the Deviator, which fires a gas cloud that converts an enemy unit to your side temporarily, and a Saboteur, and they play like an argument you are having in bad faith.

That mattered because asymmetry was rare and expensive. It is easy to ship three factions with recoloured sprites and slightly different numbers. Giving each side a unit that changes what the player is trying to do is design work, and Westwood did it in the first game of the genre, which is why Command & Conquer’s GDI-versus-Nod and StarCraft’s three-way both had a model to look at. The campaign structure helped — a map of Arrakis where you pick your next territory from a couple of options, which gave the progression a shape beyond a mission list.

Then there are the worms. A sandworm is a hazard the game runs on its own initiative: it moves under the sand, it is drawn to vibration, and it eats your harvester. It cannot be reasoned with and mostly cannot be killed, and its function is to make part of the map expensive. Rock is safe and sand is productive and dangerous, so the terrain itself is a risk calculation before the enemy has done anything. That is the sort of environmental design that went missing for twenty years while the genre argued about build orders.

The worms also solve a problem that the genre spent the next decade pretending it did not have: a map with no third party is a map where the only pressure comes from a player who might be asleep. An RTS mission where the AI turtles is dead air. A sandworm makes the ground itself impatient. Your harvester is out there on the productive sand right now, and something is coming for it regardless of what the enemy commander has decided to do this minute, and that means the map is generating tension during the exact stretches where a strategy game usually generates spreadsheet management. Westwood understood in their first attempt what the genre had to relearn from Company of Heroes and its descendants.

What it got wrong, honestly

Playing it now is instructive precisely because the rough edges show you what the next five years of the genre were actually solving.

You can only select one unit at a time. There is no drag-a-box band select — that arrived with Command & Conquer in 1995 — so moving a force of twelve tanks means twelve separate clicks, and the game’s difficulty is partly a tax on your wrist. There are no control groups. There is no waypointing or queuing. The pathfinding is dreadful: units walk into rocks, units walk into each other, units decide to take a scenic route through the enemy base while you are looking at your sidebar. The AI cheats, visibly, and the late Harkonnen missions are less a strategic problem than an endurance one.

None of that is a reason to be sniffy about it. Those are the artefacts of a team building a control scheme with no precedent to copy, and the interesting thing is how much they got right first time. The sidebar is still the sidebar. The right-click-to-move is still right-click-to-move. Westwood’s own C&C, three years later, is recognisably the same game with the ergonomics fixed, and the reason it sold in the numbers it did is that the underlying machine was already correct.

The systems lesson worth carrying out of it: a genre gets fixed by whoever solves its input problem, and the solution is stickier than any individual mechanic. Dune II’s economy has been reworked a hundred times — no harvesters, capture points, passive trickle, supply lines. The sidebar and the cursor survived all of it untouched.

Where it sits

Dune II belongs with the small group of early-90s games that shipped a complete idea rather than an iteration. Syndicate did it a year later with real-time squad tactics and a persuasion mechanic nobody has had the nerve to reprise. Cannon Fodder took the mouse-driven real-time squad down to two buttons and made it mean something. All three understood that the mouse had changed what a strategy game could ask of a player, and all three were built by people who had spent the 80s watching strategy games ask for a lot and give back very little in return.

Play it through OpenDUNE or Dune Legacy, the open-source reimplementations that keep the design and fix the pathfinding, or on DOS emulation if you want the wrist tax as shipped. Take the Atreides campaign. Watch the first harvester drive out onto open sand, and notice that you are already anxious about it, and that the anxiety is the entire genre arriving at once.

Westwood built a template so good that its greatest achievement is being invisible. Every base you have ever built, in every RTS you have ever played, is standing on concrete they poured on Arrakis in 1992.