Thrust: The Physics Game in 16K

The price is the first thing worth arguing about. Thrust arrived from Firebird in 1986 on the Silver budget range at £1.99, which in 1986 was pocket-money money — the price of a paperback, roughly a third of what a full-price C64 game cost. What Jeremy Smith put on that cassette was a working two-dimensional simulation of gravity, thrust and rotational momentum, with a rope in it. Silver was where you found the cheap tie-ins and the reheated type-in listings. It is also where the most rigorous piece of physics design on the machine turned up, at a price that implied nobody expected much.

I want to be precise about what Thrust does, because “it has physics” has become a phrase that means almost nothing. Plenty of C64 games had inertia in the sense that your sprite kept moving for a moment after you let go. Thrust has physics in the sense that the game contains no other verb. There is no jump. There is no walk. There is a ship that points somewhere, an engine that pushes it along the direction it points, and a constant downward acceleration that never turns off. Every single thing you do in the game is an argument with those three facts.

The three-button problem

The control scheme is the design. Rotate left, rotate right, thrust — and later a fire button and a tractor beam that share duties. Your ship does not steer. It aims. To move left you must rotate anticlockwise, apply engine, and then rotate back and burn against your own momentum to stop, all while gravity is quietly adding downward velocity to everything you just did. Landing is a controlled failure to crash: you have to be nearly vertical, nearly still, and descending slowly enough that the surface accepts you.

This is why the game feels so different from the shooters that surrounded it. In Uridium, Andrew Braybrook gives you a ship that responds instantly and asks you to be accurate at speed; the skill is reading the scroll and threading it. Thrust removes the instant response entirely. Your input at any moment affects the derivative of where you are, and it takes a genuine mental adjustment — most players need an hour — before your hands stop trying to fly it like a plane.

The cruelty is that the caves are narrow. Smith drew each level as a flick-free scrolling cavern with a mouth at the top and a chamber somewhere down inside it, and the corridors are frequently barely wider than the ship plus the width of your error. There is no armour and no health bar. Contact with rock is instant death. So the game is a sustained exercise in doing very small burns and living with the consequences of the burn you did four seconds ago.

The pod on a rope

Then Smith attaches the pod, and the whole thing changes register.

The objective on each planet is a Klystron Pod sitting in the cavern. You fly down to it, hold the tractor beam, and a tether snaps into being between your ship and the pod. Now you fly out. The pod is heavier than you, it hangs below you on a line of fixed length, and — this is the part that makes the game — it is a pendulum. It swings. Every burn you make transmits through the tether with a delay, sets the pod moving, and the pod then drags your ship in return. Rotate too hard and you start a swing that grows. Thrust at the wrong moment in the swing and you amplify it. The pod hits a wall and you are dead, or worse, the pod is gone.

I know of no other game from that era that models coupled masses this honestly. The tether is not decoration on top of a normal ship. It reverses the direction of authority: the pod is now steering you as much as you steer it, and the skill being tested is timing your burns to the pendulum’s period so the swing damps down instead of building. This is a real technique. You learn to burn into the swing at its extreme, bleeding energy out of it, the way you learn to stop a playground swing. The game never explains this. It just kills you until you work it out.

That is the design lesson worth taking away, and it is the same lesson Boulder Dash teaches from the other end. Rockford’s world is a grid with one rule about falling, and the whole game is emergent consequence. Thrust is continuous rather than gridded, and it does the same job: state the physical law honestly, place obstacles, and let the player generate their own difficulty curve out of their own incompetence. Both games are almost entirely free of scripted content. Both are enormous.

Fuel, limpets and the thing you should not shoot

Around this core Smith wraps a small economy that keeps the pressure on. Thrusting burns fuel. Fuel exists as pods on the surface that you refuel from by hovering over them with the tractor beam — which means the act of refuelling requires you to hold a stable low hover, the single hardest manoeuvre in the game, while your fuel is at its lowest. That is a beautifully mean loop. The moment you most need fuel is the moment you are least able to earn it.

The caverns are defended by limpet guns bolted to the rock, which track and fire, and a power plant. The power plant is the game’s best joke. You can shoot it. It takes several hits. Shoot it enough and the planet’s reactor goes critical, a countdown starts, and you have a handful of seconds to get out of the cavern and off the planet before the whole thing detonates. There is a score bonus for destroying the planet you just robbed. So the game offers you a deal: be greedy and add a hard time limit to the most delicate flying you have ever done. Nearly everyone takes the deal. Nearly everyone dies taking it.

There is also a shield, which drains fuel while it is up and functions mainly as a tax you pay for a mistake you already made.

The lineage, honestly stated

Thrust’s ancestors are in the arcade and the mainframe, which is exactly where I have no business claiming a memory. Lunar Lander put a fuel budget and a descent rate on a vector screen in 1979. Atari’s Gravitar in 1982 added the rotating ship, the gravity wells and the multi-planet structure, and if you play Gravitar today the family resemblance to Thrust is immediate — the ship, the tractor beam that pulls fuel pods, the caverns. Smith’s contribution is the tether as a physical object with mass on the end, and the decision to build the entire difficulty out of it rather than out of enemy count.

Downstream, the debt is easy to trace. Gravity Force on the Amiga in 1989 took the control scheme into two-player split-screen duels and its sequel refined it. Solar Jetman on the NES in 1990 put a tethered pod behind your ship across a set of scrolling gravity-well planets. Exile, on the BBC and later the C64 and Amiga, took the same flight model and built a whole physically simulated world around it. The line runs forward through decades of tether-physics games, most recently through everything that asks you to fly a fragile thing while something heavy hangs off it.

There’s a broader pattern here that the C64 keeps demonstrating. The machine’s best games are frequently the ones that pick a single system and refuse to add anything else. Paradroid is one mechanic — take over a robot — pursued to exhaustion. Elite is one idea about procedural generation carried to eight galaxies. Thrust is one differential equation with a rope in it, and it sold for two quid.

What it costs you now

Thrust’s reputation problem is that it looks like nothing. Wireframe-ish vector caverns, a small ship, black space. Screenshots do it no favours at all, and the budget price tag followed it into history as a slur when it was actually the most interesting thing about its release. Firebird put a genuinely rigorous piece of design at the price point reserved for filler, and a generation of British teenagers learned orbital mechanics by accident because it was the cheapest thing in the newsagent’s rack.

The C64 original is the version to seek, though the BBC, Electron, Spectrum, Amstrad and Atari 8-bit conversions all exist and the physics survives the translation better than the graphics do. Any faithful C64 emulation will do it justice; it is a game that needs a joystick and a dark room and about ninety minutes before it stops feeling arbitrary. Give it those ninety minutes. The moment the pendulum clicks — the moment you feel the swing coming and burn against it without thinking — is one of the purest things the machine has to offer.

Spoilers below

Thrust’s structure is a loop, and it is worth knowing what you’re signing up for. Clear the planets and the game does not end. It sends you back through them with the rules altered, and the alterations are vicious in a way that is genuinely funny once you stop swearing.

Reverse gravity is the first insult. The constant downward acceleration you have spent hours internalising flips, and everything you learned inverts with it: your hands know the old model and your hands are now wrong. The pod hangs above you. The pendulum still works, and your instincts about which way to burn are precisely backwards.

Then the landscape goes invisible. The caverns are still there. The collision is still lethal. You simply cannot see the rock, and your only feedback is your memory of the level plus the shape of the fuel pods and guns still drawn on the screen. It is a level-design troll, and it works only because the flight model is exact enough to fly blind on. You know where the ship goes when you burn. That is the whole reason the game earns the right to take the picture away.