The Mascot Platformer Boom and Bust

For roughly five years every publisher in the business believed it needed an animal with an attitude. Between Sonic in 1991 and the shelves clearing around 1997 we got a bobcat, a squirrel, a possum, a gecko, a bat, a bug, a crocodile, an opossum, several hedgehog adjacents, and a ninja who advertised lollipops. Almost all of them were bad. The interesting question is why the same idea, executed by competent studios with real budgets, produced a run of failure that specific.

The usual answer is that 3D came along and killed the genre. That answer is wrong on the dates, and being wrong on the dates matters, because the real reason is a design lesson that keeps recurring.

What Sonic actually was

Sonic the Hedgehog arrived in June 1991, out of a Sega team including Yuji Naka, Naoto Ohshima and Hirokazu Yasuhara, and Sega’s brief was commercially explicit: they needed something to point at Mario. Alex Kidd hadn’t worked. The character came out of a design competition. Every bit of the origin story is marketing-shaped.

The game underneath is a physics engine.

Sonic’s whole identity is momentum. You accumulate speed over distance, you conserve it across slopes, you convert it into height by hitting a ramp, and rolling turns you into a ballistic object that trades control for velocity. Yasuhara’s level design is built to teach that trade: the loops exist because a loop is a legible demonstration of centripetal force, and the game gives you a fast route and a slow route through most zones so that speed becomes a skill you’re spending rather than a cutscene you’re watching.

The character is a summary of the mechanic. He’s blue and fast and impatient because the system is about kinetic energy, and the sunglasses attitude in the marketing is downstream of a physics model that already existed. Mario’s the same relationship from the other side — that stubby run-up and the way he skids is Nintendo’s acceleration curve wearing a hat.

The thing that got copied

What the industry took from Sonic was the sunglasses.

The reading in publisher boardrooms was that Sega had won a character war, and that characters could be manufactured. So the pipeline ran backwards: commission a mascot, approve the mascot, then get a studio to build a platformer for it to be in. The design document begins with a personality and works towards a mechanic, which is precisely the order in which nothing good has ever been made.

You can hear the reversal in how those games were sold. Sonic’s marketing talks about speed. Bubsy’s marketing talks about Bubsy. The pitch is the character’s voice — he quips, he’s sarcastic, he has catchphrases — and the game is where the character is stored.

Why the boardroom wanted it anyway

It’s worth being fair to the publishers, because the logic wasn’t stupid on its own terms.

A game sells once. A character sells cartoons, lunchboxes, sequels, a comic, a licensing line, and a reason for a retailer to give you shelf space next year. Sonic gave Sega a Saturday morning cartoon and a balloon in the Macy’s parade. That is an asset class, and a platformer in 1993 cost a fraction of what an asset class is worth, so the expected value of commissioning one was enormous even at a low hit rate.

What the maths missed is that the asset only exists if the game is loved, and the game is loved because of how it moves. The character is a handle on an experience. Manufacturing the handle and hoping an experience turns up to be held is a bet on the least reliable step in the chain, and publishers made that bet dozens of times in four years because the upside was so large that the failure rate looked survivable. It was survivable, financially. It was corrosive to the genre, because it filled the shelves with games that had a personality and no verb.

Zool, and a lollipop

Zool: Ninja of the Nth Dimension, from Gremlin Graphics in 1992, is the honest case because the seams show.

It was positioned as the Amiga’s Sonic, and it moves quickly, and it has almost no momentum model — Zool is fast in the sense that his sprite has a high top speed, and speed without conservation is just a number. The physics don’t reward anything. The levels are themed around confectionery, and the confectionery is Chupa Chups, who paid for it. Level one is a sweet factory with the brand on the walls.

I don’t hold the product placement against it, particularly. What it reveals is the priority order: a game whose first world was negotiated with a sponsor is a game where the world came before the movement. Compare what Manfred Trenz was doing on the same machine with Turrican, where the level design and the traversal argue with each other productively, or where the Amiga’s platformers actually earned their reputations. Zool sold well. It also evaporated, and Gremlin never got the franchise it had budgeted for.

Bubsy, and a man who used to write Infocom games

Bubsy in Claws Encounters of the Furred Kind, Accolade, 1993, is the boom’s monument.

Accolade spent heavily. There was a television pilot. The design lead was Michael Berlyn, who had written Suspended and Infidel for Infocom — genuinely one of the more thoughtful authors in the business, working on a bobcat with a catchphrase. That’s the era in one credit line.

The game’s central problem is that Bubsy dies in one hit while being built for speed. Those two decisions are incompatible: momentum means committing to a trajectory before you can see where it lands, and one-hit death means every commitment is potentially fatal, so the correct way to play is slowly, which the game is not designed for. The mechanic and the character are pulling opposite ways, and nobody resolved it because the character wasn’t derived from the mechanic in the first place.

The levels compound it. They sprawl vertically as well as horizontally, with drops that require faith, and a fast character with one hit point exploring an unreadable space is being punished for using his defining ability. Accolade’s answer was more Bubsy — more quips, more voice clips — because the character was the product and the product was performing exactly as commissioned.

Then Bubsy 3D in 1996, which is the punchline everyone knows.

The bust wasn’t 3D

Here’s the dates problem. Super Mario 64 shipped in 1996 and is a platformer. Crash Bandicoot shipped in 1996 and is a platformer with a Sony-manufactured mascot on the box. Spyro followed in 1998. Banjo-Kazooie in 1998. The genre was doing fine in three dimensions, and Sony ran the mascot playbook deliberately and it worked.

What died was the manufactured mascot with no system under it, and it died because players had, by about 1995, played eleven of them and developed a nose for it. Gex arrived in 1995 with a lot of voice acting and a fairly ordinary game. Bug! on the Saturn had a genuinely novel idea — traversal along tracks in 3D space — and shipped stiff. Awesome Possum spent its budget on a theme tune with lyrics.

Crash worked because Naughty Dog started from a technical thesis about what the PlayStation could draw and built the character to fit it. The mascot came last. Same order as Sonic.

The ones that survived, and why

The survivors of the era all have a mechanic you can name in a sentence.

Rocket Knight Adventures, Konami, 1993: Sparkster has a jetpack with a charge meter that turns him into a projectile you aim, so the game is about committing to a vector. Earthworm Jim, Shiny, 1994: Doug TenNapel’s design uses the worm’s own body as a whip and a grapple, and the humour lands because the game is in on the joke about mascots. Cool Spot, from David Perry’s team at Virgin in 1993, is an advert — the 7 Up mascot, contract work — and it plays well because the movement and the shooting were designed properly by people who took the assignment seriously.

Even on the ageing 8-bit machines you can watch the same rule hold. Mayhem in Monsterland on the C64 in 1993, from the Rowland brothers, chased the Sonic feeling on hardware that had no business producing it, and it works because the whole thing is organised around one mechanic: colour returns to the world as you progress, and the speed is earned through the level rather than granted. The C64 was capable of that kind of trick when someone started from the system.

The lesson isn’t complicated and the industry relearns it every cycle. A character can summarise a mechanic beautifully and become an asset worth a fortune. The summary has to have something to summarise. Manufacture the summary first and you get a bobcat with a catchphrase, an incompatible death rule, and a five-year hole in a publisher’s slate — a pattern that also explains most of what the licensed tie-in era produced, for the same reason, starting from the same end.