Turrican: The Sprawl That Broke the Rules

Most C64 platformers of the late eighties were built to be learned. The screen scrolled one way, the enemies came on a fixed schedule, and the skill on offer was the skill of a pianist: play the same bar enough times and your hands stop asking your brain for permission. Turrican, which Rainbow Arts put out in 1990, declined that whole arrangement. Its levels are large enough that memorising them is a project rather than a session, they scroll in every direction, and several of the best things in them are behind walls that give no indication of being anything other than walls. It is a game about occupying a space instead of completing a corridor, and on a machine with 64 kilobytes of RAM that was an odd, expensive, slightly mad thing to attempt.

The programmer who lost a lawsuit and won a career

Manfred Trenz’s route to Turrican runs through two of the most entertaining legal near-misses in European games. He was the artist on The Great Giana Sisters, the 1987 Rainbow Arts platformer whose resemblance to a certain Nintendo product was close enough that it left shelves in a hurry. Then he programmed Katakis in 1988, a horizontal shooter that resembled R-Type the way Giana had resembled Mario. Activision held the home-computer rights to R-Type and objected; Katakis was pulled. The sequel to that story is the good part: Rainbow Arts ended up with the contract to convert the actual R-Type, and Trenz wrote the C64 version himself. A man who had been legally instructed to stop imitating a game was then paid to reproduce it exactly.

You can feel both halves of that education in Turrican. The imitation instinct is still there — Trenz has been open about Metroid and the arcade game Psycho-Nics Oscar as sources, and the beam weapon comes straight out of the latter. What’s new is the confidence of somebody who has now shipped a conversion of one of the most demanding arcade games in existence on an eight-bit machine and knows exactly what the hardware will tolerate. Turrican spends that knowledge on size.

The map is the mechanic

Five worlds, cut into thirteen levels, and the first thing the first level does is scroll upwards when you expected it to scroll right. That sounds trivial now. In 1990 on a C64 it reset your posture. A left-to-right platformer tells you where the game is: the game is over there, past the next enemy. Turrican declines to tell you. There are branches that rejoin, branches that don’t, vertical shafts that turn out to be the main route, and side pockets stuffed with extra lives that exist only for people who went looking.

The design consequence is that Turrican converts the player’s attention from a timing problem into a spatial one. In a scrolling shooter the question is when. In Turrican the question is where, and where is a question you can be wrong about for two minutes without dying, which is a strange and rather generous kind of failure for an eight-bit action game to offer. It also produces the thing the game is most criticised for and most loved for at once: you will wander. Some of that wandering is the game working. Some of it is the game being enormous without always being legible, and Turrican has no map, no compass, no marker. It has your memory and a screen that keeps moving.

What holds it together is that the traversal itself is pleasant. Turrican’s sprite has weight and speed at the same time — he accelerates to a genuine sprint, the jump has a real arc, and the collision is honest. That matters more than any level layout. A big level in a game with bad movement is a punishment. A big level in a game with good movement is a place. Rainbow Arts got the movement right first, and everything else in the design is downstream of that decision.

The beam, the wheel, the mines

The beam is the reason Turrican looks like nothing else on the machine. Hold fire and Turrican extends a whip of energy that you rotate a full 360 degrees with the joystick, sweeping it around like a torch in a dark room. It kills things, which is the obvious use. The real use is that it lets you interrogate the level. You stand still, you sweep the beam across a wall, and if the wall is a lie the beam finds the lie for you.

Think about what that does to the loop. Every other game of this vintage separates exploring and shooting: you move through the level, and when a thing appears you shoot the thing. Turrican welds them together. The weapon is a search tool. The act of poking at scenery — normally the most tedious thing a game can ask of you — becomes a physical action with a satisfying analogue input, because rotating the beam with a joystick feels good in the hand in a way that pressing a “search” button never has. This is a design idea a lot of modern games are still fumbling. Turrican solved it by accident of theme in 1990.

Around the beam sit the other verbs. Three weapon lines you swap by collecting tokens, a screen-clearing lightning attack held in reserve, and the wheel: press down and Turrican folds into an invulnerable rolling sphere that lays mines and fits through gaps his standing frame cannot. The wheel is the clever one, because it is not a power-up. It is a shape, available always, with a cost — you can’t shoot, you can’t jump properly, and the energy drains. So the wheel turns awkward geometry into a decision. A low gap with something nasty behind it is a question: do I roll through blind and armed with mines only, or do I go the long way? The game asks that question dozens of times and never announces that it is asking.

What it costs

Turrican is not kind. Enemy placement is frequently a matter of walking into a room and discovering, at the speed of a C64 sprite, that the room contained something. Because the levels are big and the checkpointing is sparse, a bad death costs real ground. And the sheer scale that makes the game special also makes it baggy: there are stretches of world three and four that are large without being interesting, corridors that exist to be long. Trenz was building at the edge of what the machine could hold, and the seams show where the ambition outran the content.

The soundtrack does a lot of load-bearing work here. Chris Huelsbeck’s score is one of the SID chip’s high-water marks, and it functions as level design: the music gives a sprawling, direction-free space an emotional shape it hasn’t earned geometrically. Walk around lost for ninety seconds with that title theme playing and you feel like an explorer. Walk around lost in silence and you feel like someone who took a wrong turn. Huelsbeck is why the wandering reads as atmosphere.

The economics of a big level in 1990

It’s worth being concrete about why nobody else did this. A C64 has 64K of RAM and Turrican shipped on tape and disk in a market where the cassette was still the default in Britain and Germany. A level the size of Turrican’s cannot be stored as a picture; it has to be stored as a recipe. The whole game is built from small character blocks assembled into larger tiles, those tiles assembled into screens, and the screens into worlds — compression through repetition, with the art directed specifically so that repetition reads as architecture instead of wallpaper. Trenz was drawing and coding at once, which is why it works: the man deciding what a wall looked like was the same man who knew how many bytes a wall could cost.

That constraint shaped the play as much as any design document. The reason Turrican’s worlds have a texture — caverns that feel like caverns, factory levels that feel assembled — is that the tile vocabulary is small enough to become a visual grammar. You start recognising blocks. And once you recognise blocks, you start reading the level the way a climber reads a rock face, noticing that this ledge arrangement usually means there’s something above it. The game teaches you its own dialect without a word of tutorial, purely because it could not afford to say anything twice.

Modern procedural generation chases the same effect and mostly loses it, because a system that can say anything says nothing memorable. Turrican had a hard ceiling on vocabulary and turned the ceiling into style.

Where it went

The obvious descendant is Turrican II, which took the same architecture to the Amiga a year later and had the hardware to make the sprawl look as good as it played. The less obvious descendants are everywhere. Every time a modern action game hands you a large, unsignposted space and expects the movement itself to be worth the walk, it is making Turrican’s bet: that a good verb set turns navigation into content. Most of them hedge that bet with a minimap. Turrican didn’t hedge anything.

Its C64 contemporaries mostly went the other way, and they were right to. The great C64 action games are usually exercises in compression — Uridium is a single idea sharpened until it draws blood, Armalyte is a shooter that treats every screen as a machined component. Turrican is the opposite instinct on the same hardware: spend the memory on room. Both philosophies produced masterpieces on the same machine within a couple of years of each other, which tells you something about how much design space the C64 still had left in 1990. Mayhem in Monsterland would prove there was more still, three years after Turrican and one year before Commodore’s collapse.

Play it on real hardware or a decent emulator with a joystick that has a microswitch in it, because the beam is an analogue gesture and a keyboard murders it. Give it an hour before you decide. The first twenty minutes of Turrican feel like being handed a map of a city you have no reason to visit. The hour after that is when you notice you’ve stopped following the level and started reading it, and that switch — from execution to attention — is what the whole enormous, uneven, brilliant thing was built to produce.