Bubble Bobble on the C64: The Conversion That Held Up

The arcade conversion was the great British disappointment of the 1980s, and every one of us walked into it repeatedly with our eyes open. You played a cabinet in a chip shop, the cabinet had custom sprite hardware and a couple of dedicated sound chips and cost more than a car, and then eighteen months later a company bought the licence and paid one programmer to reproduce it on a machine with one megahertz and a single graphics chip inside a six-week schedule. The box art was the arcade artwork. The screenshot on the back was chosen with care. The tape inside was an apology.

Out Run is the standard example and it earned the position. The C64 version, from US Gold in 1987, shipped as the biggest licence of the year with a Ferrari on the cover and a frame rate that made you seasick, and the industry’s own magazines said so at the time. The pattern repeated for years. Which is why Bubble Bobble matters: Firebird put it out in 1987, on the same machine, in the same market, under the same constraints, and it works.

What Taito actually built

Fukio Mitsuji designed Bubble Bobble for Taito’s arcades in 1986, and the design is a small, tight machine with one verb.

You are a dragon. You blow a bubble. The bubble drifts. If it hits an enemy the enemy is now inside it, floating, alive and furious and temporarily harmless. You can jump on that bubble and ride it. You can burst it, at which point the enemy is gone and a piece of fruit drops. Or you can leave it, gather three or four more trapped enemies in bubbles, and burst them in a chain — which multiplies the score and drops much better items.

Every interesting decision in the game comes from that delay between trapping and killing. A trapped enemy is a resource with a timer on it: leave the bubble too long and it pops on its own and the enemy comes back angrier and faster. So each room is a negotiation with your own greed. Clear the room immediately and be safe and poor, or leave four bubbles bobbing while you hunt a fifth and be rich or dead. The whole game is that dial and the player sets it themselves, moment to moment, a hundred times over.

The other half is the bubble as terrain. Your bubbles rise, gather at the ceiling, drift along the tops of platforms, and you can stand on them. So the level geometry is partly something you have manufactured, and a room you cannot reach the top of becomes a room you build a staircase in. Mitsuji got two systems out of one button, and this is why the game has never dated: the mechanic generates its own complications faster than a designer could author them.

The nearest relative on the C64 shelf is Boulder Dash, which does the same trick from a different angle — one physical rule, applied without exception, producing puzzles the authors did not have to write. Both games are proof that a mechanic with genuine internal logic outperforms a mechanic with content bolted to it.

Why the conversion survives

Firebird’s C64 version, released in 1987, makes a set of choices that most of its contemporaries got wrong.

It keeps the complete structure. All hundred rooms. Two-player simultaneous co-operative play, which is the mode the game was designed around and the mode most ports of the period quietly dropped for a “player one, then player two” alternation because simultaneous sprites were expensive. It keeps the bubble physics, which is the only thing that actually had to survive — the drift, the ceiling gather, the ride. It keeps the potion power-ups and the E-X-T-E-N-D letter bubbles that grant a free life when you collect the set.

What it gives up is the arcade’s visual density and some of its speed. The sprites are chunkier. There is flicker when the room fills. Colour clash in the C64 sense — the multicolour mode’s shared-palette restriction — flattens things.

And none of that matters, because the thing being converted was a loop, and the port’s author understood which parts of the loop were load-bearing. This is the skill that separated the good conversions from the bad ones, and it was never primarily a technical skill. Chris Butler’s Commando on the C64 is celebrated for the same reason: he worked out what the arcade game was for and rebuilt that, accepting losses everywhere else. The Out Run port failed because the thing being converted was a sensation of speed, and speed was precisely the thing the hardware could not give, and nobody stopped to ask whether a different set of compromises might have delivered the sensation another way.

Bubble Bobble had an easier brief, in fairness. A game about bubbles rising slowly is a game whose core does not require frames. There’s a real lesson in that: the arcade titles that converted well are almost all the ones whose identity lived in a rule rather than in a rush. Impossible Mission was born on the machine and had the same luxury. The C64 was a superb rules computer and a poor spectacle computer, and the publishers who bought licences kept buying spectacle.

Firebird deserves some of the credit here, and it is an odd company to be handing credit to. It was British Telecom’s software label — a nationalised telecoms monopoly selling cassettes, which is a sentence that only makes sense in the context of 1980s Britain — and it spent most of its life on the budget Silver range at £1.99, where it also published Thrust and the C64 conversion of Elite. A label with that catalogue plainly had somebody in it who could tell the difference between a game and a licence. Bubble Bobble was a full-price release and the money shows in the room count: a hundred screens is a lot of data to fit on a cassette, and the temptation to ship sixty and hope nobody counted was real and routine and other people took it.

The co-op is the argument

Two players is where Bubble Bobble becomes the thing people remember, and it is worth being precise about why, because “it’s fun with a friend” is the laziest sentence in games writing.

The mechanic creates interference. Your bubbles are terrain for your friend. Your friend’s bubbles are terrain for you. A bubble you blew to trap an enemy is a platform they are now riding into danger, and an enemy you were saving for a chain is one they just burst for the small score. You can trap each other’s targets. You can shove each other around by bubbling into them. So co-operation is constantly being undermined by the fact that the same object serves both of you differently, which produces the specific texture of Bubble Bobble arguments: it is impossible to tell whether your friend has just helped you or killed you, and often it was both.

Very few co-op designs from any decade get this. The default modern approach is to give each player their own lane and score, which removes friction and removes the game. Mitsuji made one shared physical space with objects that belong to nobody, and forty years later Deep Rock Galactic is still working the same seam — shared terrain that any player can alter and everyone has to live in.

The C64 version keeps this intact, on one machine, two joysticks, one sofa. That was the entire point of the purchase.

Where to play it

The arcade original is widely available in Taito collections and it is the definitive version; play it if you want to see what Mitsuji intended, at the speed he intended it. The C64 conversion is the one worth playing if you want to see what a conversion could be when someone thought about it — and it is genuinely playable now, which is a sentence I would not write about the majority of its shelf-mates from 1987.

Play it with somebody. Alone it is a good game about greed management. With a second joystick it is a machine for generating grievances, and the game knows this and rewards you for it in ways the single-player mode explicitly refuses to.

Spoilers below

The reward for co-operation is structural, and the game tells you about it to your face.

Finish all hundred rooms on your own and the game gives you an ending that announces itself as false. You are shown a message telling you, plainly, that the true ending exists and that it requires two players. This is a game from 1986 withholding its real conclusion and printing the reason on the screen. It is outrageous and it is the single most confident piece of design in it.

With two players you get the actual final sequence: the Super Drunk boss on the hundredth floor, defeated by bubbles, and Bub and Bob restored to the human boys they were before the game started. The dragons were the curse. The whole cave was a spell, and the spell only breaks for two people working together, because the game has been a hundred rooms of argument about exactly that. Mitsuji built a 100-level tutorial for its own ending.

There are further layers under that — secret rooms, hidden doors, the fabled conditions for the alternative endings — and the C64 version carries a good deal of it across. Forty years of documentation exists. I would leave it undocumented for one playthrough and let the game withhold something from you honestly. It earned the right.