The C64 Canon: Fifteen Games That Still Hold Up

Every retro canon has the same failure mode: it lists the games the writer loved at eleven and calls the love a judgement. I had a C64 from about 1984, so I’m exactly the person that failure mode is built for, and the only defence I’ve found is a hard test. Does the system still work? Strip out the affection, the tape hiss, the smell of the carpet in front of the telly. Hand the design to someone born in 2005. Does it hold?

Fifteen that do. The shooters are thin on the ground here on purpose — they get their own canon, because the C64 produced enough great ones to make this list into a genre exercise if I let it.

The early strategists (1983)

M.U.L.E. (Ozark Softscape / Electronic Arts, 1983). Danielle Bunten Berry built a four-player economics game in which the auction is the entire drama. Players trade food, energy, smithore and crystite on a colony world, and the market clears through a real-time double auction where you physically walk your avatar up and down a price axis. The genius is the incentive geometry: the game scores the colony collectively and ranks you individually, so you need your rivals to survive and want them to finish second. That tension is still ahead of most modern multiplayer economies. It plays today with no apology required.

Archon (Free Fall Associates / Electronic Arts, 1983). Chess where taking a square starts an action fight, and the board’s light and dark squares cycle through a day-night phase that buffs whichever side matches. Two systems that should reject each other — turn-based positioning and twitch combat — bolt together because the fight inherits the strategy: a weakened piece stays weakened. The full argument is here.

The 1984 pair

Impossible Mission (Epyx, 1984). Dennis Caswell’s game is a search problem wearing a platformer. You raid rooms for puzzle pieces, assemble a password, and the somersault — the animation everyone remembers — is a movement verb with real commitment cost. The speech synthesis on a 1984 C64 (“Another visitor! Stay a while… stay forever!”) is the famous part. The durable part is that the game respects your time in a way its contemporaries didn’t: search is skippable with lift puzzles, and the rooms are generated per playthrough. More on why the somersault matters.

Boulder Dash (First Star Software, 1984). Peter Liepa and Chris Gray shipped a physics grid so clean it became a genre. Rocks fall, dirt supports, diamonds roll off curved surfaces, and every cave is a proof about gravity you solve by digging in the right order. There’s no fat on it at all — the entire game is one rule applied honestly. The grid that taught physics.

The 1985–86 peak

Elite (Firebird, C64 version 1985). Bell and Braben’s universe compressed into procedural generation because there was no room for a universe. Eight galaxies, 256 planets each, generated from a seed — the constraint invented the technique that every open world now uses. Trading, fighting and the terrifying business of docking without an autopilot. The universe in 32K.

Paradroid (Hewson, 1985). Andrew Braybrook’s masterpiece, and the tightest system on the machine. You’re a droid that takes over other droids via a transfer minigame — a little circuit-flow duel — and the hierarchy of enemy droids is a legible power curve you climb by winning arguments with electricity. The minigame is the progression system, which is a design economy almost nobody manages. Braybrook’s perfect little system.

Little Computer People (Activision, 1985). A man lives in your computer. He has a dog. He writes you letters, plays the piano, gets depressed if you ignore him. There’s barely a game here and it doesn’t matter — this is the ancestor of every simulation about tending a thing that has its own opinions, fifteen years before The Sims. The Sims before The Sims.

The Sentinel (Firebird, 1986). Geoff Crammond’s game has no relatives. You absorb objects for energy, create a robot body on higher ground, transfer into it, and climb a landscape towards a rotating Sentinel that drains you if it can see you. 10,000 landscapes from a seed. It’s a game about line of sight and elevation as the only two resources, and it remains genuinely alien to play. The strangest game on the C64.

Uridium (Hewson, 1986). Braybrook again, and the one shooter I’m keeping here, because it isn’t really a shooter — it’s a flight model. Fifteen dreadnoughts, a Manta that turns with weight and momentum, and hull architecture that kills you more often than the enemies do. The shoot-em-up as precision instrument.

The 1987 class

Wizball (Sensible Software / Ocean, 1987). Jon Hare and Chris Yates made a game about restoring colour to a monochrome world by collecting paint droplets with a cat. The control is the joke and the point: you start as a bouncing ball with no steering and buy your own controllability with upgrades, so mastery is literally purchased. The mission-structure-inside-an-arcade-shooter idea landed thirty years before it was fashionable. The colour-restoring oddity.

Pirates! (MicroProse, 1987). Sid Meier’s Caribbean sandbox was a C64 original, and it invented the open-world structure people credit to games a decade younger: no fail state worth the name, a clock that ages you out, and four or five minigames stitched into a career you narrate yourself. You can just sail off and do nothing. In 1987 that was close to heresy. The freedom Sid Meier found first.

IK+ (System 3, 1987). Archer MacLean put three fighters on screen on a C64 and made the third one a comedian. The fighting is precise — a real move-priority system with recovery frames — and the game keeps undercutting its own seriousness with the little man who wanders on and drops his trousers. Fighting games got more complex afterwards; very few got funnier. The fighting game as comedy.

The late machine

Last Ninja 2 (System 3, 1988). The isometric one everyone means when they say the C64 looked incredible. Manhattan rendered in a projection that shouldn’t have been possible on the hardware, with combat that’s stiff and puzzles that are cruel — and it still holds because the place holds. The isometric peak, and the wider isometric canon.

Turrican (Rainbow Arts, 1990). Manfred Trenz’s answer to the question nobody asked: what if a run-and-gun level was the size of a small country? Enormous branching maps, a beam weapon you steer with the stick, and a gyroscope mode for the tight bits. The sprawl broke the genre’s rules about pacing and got away with it. The sprawl that broke the rules.

Mayhem in Monsterland (Apex Computer Productions, 1993). The Rowlands brothers shipped this a year before Commodore discontinued the machine, and it does things — full-screen smooth scrolling, a colour palette that reads as bright rather than muddy — that the C64 was widely agreed to be incapable of. Commodore Format gave it 100%, the only time the magazine ever awarded that, and Apex sold it largely themselves to cut out the middleman. The C64 doing the impossible.

What the fifteen have in common

Read the list back and a pattern shows up. Almost every entry is one idea taken seriously — a transfer minigame, a gravity rule, a line-of-sight resource, an auction — rather than a bundle of features. That’s partly virtue and mostly constraint: 64 kilobytes and one programmer working for a few months produces focus by force. You can hear the same logic in the budget labels, where £1.99 bought a single mechanic done well.

The other thing they share is that none of them explain themselves. No tutorial, no quest marker, no tooltip. You worked it out from the inlay card and a friend’s rumour, which was often wrong. Some of that was cruelty. Some of it was the art of not explaining, and the games above are the reason I still think the modern default errs too far the other way.

Where to play them: most of these are legally available through the various C64 collections and mini-console releases, and the rest are emulator territory. The machine is dead and the systems aren’t.