Why 16-Bit Sports Games Were Better Systems

I want to narrow the claim before I make it, because the wide version is nonsense.

EA’s football games contain more actual football than anything on an Amiga ever did. The ball physics are genuinely modelled. The players are photogrammetry of real bodies. The animation blending is a decade of solved problems most studios can’t afford. As simulations of the televised sport, they are better by every measure that can be measured.

As systems — as machines you operate, read, and get better at in a way you can articulate — the 1990-94 stuff on the Amiga is still ahead, and the reason is structural rather than sentimental. Those games couldn’t afford a presentation layer, so every resource went into the model, and the model had to be legible or the game was unplayable. That’s a constraint that happens to produce good design, and it went away.

Small sprites are an information decision

Sensible Software’s house style gets filed under charm. The players in Sensible Soccer are about sixteen pixels tall and have no faces. It looks like a joke and it is a doctrine, which Jon Hare and Chris Yates applied to everything they made.

Here’s the reasoning. In a top-down football game, the thing you need is the pass before you make it — where the run is, where the space is opening, which defender is out of position. That information lives in the distance between things. Zoom in for detail and you buy nicer players and lose the pitch. Zoom out and the players become dots and the geometry becomes readable.

Sensible zoomed out. Sensible World of Soccer in 1994 shows you enough of the field, plus a radar for the rest, that you are playing the shape of the game rather than the man on the ball. Which is what football actually is, tactically, and which is almost exactly the view a manager has and a television camera does not. The sprite size is a decision about what the player is allowed to know.

Modern sports games have inherited the broadcast camera, and the broadcast camera exists to make football watchable on television by people who want to see faces. It is a poor instrument for playing football, so the games bolt a radar into the corner and expect you to read a map while looking at a stadium. The information architecture is fighting itself, and it’s fighting itself because the visual style was decided by marketing requirements that the 16-bit games never had to satisfy.

The ball that wasn’t yours

Dino Dini’s Kick Off, from 1989, made one decision that reorganised the genre: the ball is a separate object with its own physics, and it is never attached to your feet.

Dribbling in Kick Off means repeatedly catching up with a ball you have knocked forward. Turn too sharply and it’s gone. This was widely described as difficult and it is difficult, and it is also the single most honest thing a football game has ever done, because it puts the actual problem of football — controlling an object that has momentum and does not want to stay with you — into the input layer where the player has to solve it.

Sensible went softer, with closer control and aftertouch: bend the ball after you’ve hit it by holding a direction, which is a bit of arcade generosity that gives you an expressive skill ceiling nobody teaches you about. Microprose Soccer had done a version of this in 1988 with banana shots. Both approaches share a principle — the ball’s state is readable at all times and your input maps to it directly, with nothing in between.

The two-player consequence is the part that gets forgotten. Kick Off and Sensible Soccer were played on one machine with two joysticks and a shared screen, and a shared screen means neither player has hidden information. Everything either of you knows, you know from the same pitch. That is a symmetry modern sports games mostly cannot offer — split-screen or online, each player has their own camera, their own occlusion, their own radar. The competitive integrity of a 1992 Amiga football match was underwritten by there being exactly one view of the world, and the arguments afterwards were about skill rather than about what either of you could see.

What’s in between now is animation. Modern sports games are animation-driven: your input requests a state, an animation system decides which clip can transition from the clip currently playing, and the outcome arrives some frames later. This makes the game look like sport and it means the player is negotiating with a mediator. Every complaint about “input lag” and “the game decided I was going to do that” is a description of this architecture working as designed. Kick Off had no mediator. The ball was where the ball was.

Speedball 2 and the invented sport as a clean room

The Bitmap Brothers solved the problem by refusing the licence entirely.

Speedball 2: Brutal Deluxe is a made-up sport from 1990, which means nobody could complain that it was unrealistic and the designers owed no fidelity to anything. So the whole game is mechanism. Scoring routes have different point values. You can bounce the ball off the walls. You can injure the opposition and remove them for the rest of the match, and this is a strategy rather than an infraction. Between matches you spend earned tokens on your team’s attributes, so the league is a progression curve you can see the numbers of.

Every one of those is a system that a real-sport licence would have vetoed. Speedball 2 is the argument in its pure form: with no obligation to resemble anything, the Bitmaps built a thing where every element does work, and it plays better than most licensed sport thirty-five years later. Their whole catalogue runs on that willingness to design the fiction around the mechanism.

Archer Maclean got at the same freedom from another direction — IK+ in 1987 put three fighters on screen at once, which no real karate bout does, because three-way scoring is a more interesting system than two-way scoring.

The database was a system too

Sensible World of Soccer shipped with something like 1,500 clubs and tens of thousands of players, and a twenty-season career, on a machine with a megabyte of RAM. The obvious reading is that this is a feat of compression, and it is. The more useful reading is what the compression forced.

Each player is a handful of numbers — position, a few attributes, a value. That’s all there was room for. So scouting in SWOS is an activity you can actually perform: you look at a list, you compare figures, you form a theory about whether a cheap winger from a Finnish second division is worth a punt, and then the match tells you whether your theory was right, because the attributes map onto behaviour you can see at sixteen pixels tall. The loop closes. Theory, purchase, evidence.

A modern career mode has vastly richer player models and the loop does not close, because the number of variables between your scouting decision and the outcome on the pitch is larger than a person can hold. You are trusting a simulation. In SWOS you were reading one, and reading beats trusting every time you want a player to feel responsible for anything.

What the presentation layer eats

The mechanism here is money, and it’s worth being blunt about it.

A modern sports game’s budget is dominated by things that are conditions of the licence. Real player likenesses. Real kits. Real stadiums. Commentary recorded with real broadcasters. Annualisation, because the licence is annual and the roster changes. None of that budget reaches the model, and all of it has to be redone every twelve months, which also means the model can’t be rebuilt — rewriting the thing everyone plays is a risk you cannot take when the ship date is contractually fixed and non-negotiable.

John Madden Football is instructive as the counter-example, because Madden himself reportedly refused to put his name on anything with fewer than eleven players a side, and EA’s early Genesis versions consequently had to make the playbook the game. The playbook is a system: information asymmetry, bluffing, counters to counters. It survives because it was designed before the presentation layer existed to compete with it.

Where the old ones lose, honestly

They lose on the sport. Kick Off’s players cannot really turn. Sensible’s off-the-ball AI is thin. Nobody in a 1994 football game is making a decoy run because the AI has a model of intent — they’re following a positional rule. There is no context in the tackles, no body weight, no fatigue that means anything. If what you want is football, the modern ones have it and the old ones have a diagram of it.

The diagram is the point, though. A system you can hold in your head is a system you can get deliberately better at, and everyone who played Sensible Soccer seriously can tell you what they learned — which is a sentence very few people can complete about a recent sports release. Legibility was forced on those games by a lack of pixels. It turned out to be the most valuable thing they had, and the moment the pixels arrived, it was the first thing spent.