2026-09-24 · games · design · economics · philosophy · criticism

the dimension is the genre

Entry one of three in “the dimension problem.” Part two applies this to the remake business; part three spends all of it on Final Fantasy VII.

For thirty years the industry has treated the third axis as a settings slider. Better hardware, more polygons, and eventually the game you were already making, but in a volume. That framing is wrong in a specific and very expensive way. 2D and 3D are not two fidelity levels of one medium. They are two media. Moving a game across the axis is not an upgrade. It is a genre change that nobody discloses to the customer.

I got this from Sean Malstrom. He has been arguing some version of it on his blog since the Wii era — that the stampede into 3D was a strategic error, that Nintendo’s 2D lines kept posting numbers the 3D lines couldn’t while receiving a fraction of the budget, that the institutional fixation on 3D quietly let the 2D franchises rot. He is not the only person to have noticed and I don’t follow him to all of his conclusions. But he put the thought in my head, and the right thing to do with a borrowed idea is say where you got it.

What the argument has never had is a vocabulary. Without one it stays an opinion you can have at someone. This post builds the vocabulary.

the shape of the mistake

Gilbert Ryle’s name for this kind of error is a category mistake. His example in The Concept of Mind is a visitor walked through the colleges, the libraries, the laboratories and the playing fields, who then asks to be shown the University — having taken a thing of one logical type for a member of a set it was never in.

“3D is better than 2D” has that shape. It is not a false claim. It is not the kind of claim that can be true, because the two things aren’t ranked on a shared scale. Ask which is better, a novel or a photograph, and you get the same silence.

The industry did not hear silence. It heard a business plan.

Lessing already did this, in 1766

Gotthold Ephraim Lessing wrote Laocoön: An Essay on the Limits of Painting and Poetry to attack a piece of received wisdom — Horace’s ut pictura poesis, as in painting so in poetry, the idea that the arts share a set of rules and differ only in their materials. Lessing’s answer was that they don’t share rules at all. Painting arranges bodies in space and can only hold one arrested moment. Poetry moves through actions in time and can hold succession. Each art has a proper subject. Each fails when it tries to do the other’s job.

He was not building a hierarchy. He was drawing a border. The modern idea of an artistic medium — that the material constrains what the form can honestly do — essentially starts here.

Nobody thinks a poem is a low-resolution painting. Everybody in games spent three decades acting like a sprite was a low-resolution polygon.

rendering is not dimension

Two independent variables got welded into one number, and almost every bad decision downstream traces back to that weld.

Presentation dimension — how the thing is drawn. Sprites, polygons, hand-painted backdrops, HD-2D, whatever the hardware affords this year.

Ludic dimension — how many axes the player has to reason in.

These have nothing to do with each other. Street Fighter IV is polygonal models on a two-dimensional plane; the fight is a 2D fight and the audience for a 2D fight bought it. Octopath Traveler and the Dragon Quest HD-2D remakes run modern lighting, depth of field and particle work over a flat grid. Mario Wonder is a 2D game that looks like nothing that has ever existed.

You can buy every ounce of graphical modernity without buying the third axis. For a decade the platform holders, the middleware vendors and the trade press sold the two as a bundle, and studios bought the bundle, and a lot of studios are gone.

the axis test

Four things change when you add the axis. Naming them is most of the work.

1. The frame becomes a camera. A 2D game has no camera, it has a frame, and the screen is the state. Everything relevant is present and the player’s information problem is solved before they start. In 3D the camera is a second player you are required to operate, it can occlude, and it can lie. Every failure gets ambiguous: did I miss the jump, or did the camera swing?

2. Composition becomes furnishing. A 2D screen is composed. Every tile sat under somebody’s cursor. A 3D volume is furnished — most of its surface area is material rather than decision, because a person cannot hand-place a cubic kilometer.

3. Design surface inverts against production surface. A 2D game is mostly decisions with a little material attached. A 3D game is mostly material with a little decision buried in it. Malstrom’s observation that a studio full of artists will idle through a 2D project is the org-chart shadow of this. The ratio determines what the studio is, and then the studio defends the ratio.

4. Input goes from terminal to revisable. A 2D jump commits at takeoff; the arc is determined, legible, and yours to have judged correctly or not. 3D movement is continuously corrected on the way. That is a different skill, practiced by a different person, for different reasons.

the verb test

Any theory that calls Super Mario 64, Ocarina of Time, Metroid Prime and GTA III mistakes is worthless. Those are four of the best things the medium has produced and all four crossed the axis. So the theory has to predict them too.

It does, with one question. What is the core verb, and is it spatial or compositional?

Spatial verbs — move, aim, look, drive, hide, get lost — gain from the axis, because the axis is literally the thing they operate on. Mario’s verb is move. Metroid’s verb is look at a place until you understand it. Give those a volume and you have given them more of what they already were.

Compositional verbs — read the arrangement, solve the arrangement — lose, because the axis takes away the arrangement. Bomberman’s verb is reason about a grid; Act Zero put the grid inside a volume and the reasoning evaporated, leaving a man in armor in a brown room. Golden Axe’s verb is control a lane of enemies; Beast Rider dissolved the lane and shipped a camera you fight alongside the orcs. Contra’s verb is clear a screen you can see all of.

And here is the part the industry got backwards. Mario 64 did not replace 2D Mario. It founded a second line, and both lines kept selling.

Same order of magnitude across the axis. Wildly different cost structures underneath. That is not a succession, it is two products — and one of them has much better margins than the other. Nintendo booked the cheap one as the budget line anyway.

the scale floor

Clayton Christensen’s term for the next part is performance oversupply. Once a product’s trajectory of improvement passes what mainstream customers can actually absorb, the basis of competition shifts to price, convenience and reliability, and the incumbent optimizing for raw performance walks straight off the edge of the market it thought it owned. Malstrom showed up in gaming carrying The Innovator’s Dilemma, which is the other half of what he should be credited with.

Visual fidelity passed that line a while ago. The budgets did not notice. Industry estimates put average game budgets around $18–24 million in 2010 and $150–300 million by 2024 — roughly eightfold in fourteen years — with art production running a quarter to a third of the total and still climbing double digits year over year.

Some of that is Baumol. A string quartet still needs four players and the same twenty minutes it needed in 1800; the productivity gains happening everywhere else show up as rising costs here rather than rising output. Authoring a believable room is closer to the quartet than to the factory. Engines got better. The hours to make a room worth standing in did not fall proportionally.

Which produces the concept that does the most work in this whole argument. The third axis raises the minimum viable budget for an entry in a category. Call it the scale floor. When the floor rises past what a category’s audience can repay, that category is evicted from the platform. Not out-competed. Not rejected. Priced out. This is why 2D platformers spent a decade exiled to handhelds, then to Steam — the handheld had a lower floor, and Steam had no floor at all.

It also means the market never actually voted. Nobody asked for the third axis. They were offered products that had already paid for it.

what they were buying instead

Carl Menger’s move in 1871 was to stop looking for value inside the object. Value is not a property a thing carries around; it is a relation between a thing and a particular person’s particular want, at the margin, right now. Nothing has value in itself, including a third axis.

Line it up against the verb test and the marginal utility of the axis sorts cleanly. Positive where the verb is spatial. Near zero where the verb is compositional. Negative the moment it costs legibility — because for a compositional game, legibility is not a feature, it is the product.

Silent Hill’s fog is the sharpest case of this and part two will spend it properly. The fog was a hardware limitation turned into the entire aesthetic. The HD re-release resolved the scene and removed the fog, and shipped a game with more pixels and less of what anyone was buying.

Receipts on the other side, all of them from products that stayed in their category:

• Dragon Quest III HD-2D plus I & II HD-2D, past 4 million combined
• Octopath Traveler series, past 7 million
• Streets of Rage 4, past 2.5 million
• Sonic Mania, past 1 million
• TMNT: Shredder’s Revenge, 1 million in its first week

None of these cost $200 million, because none of them had to. Their floor was low, so their ceiling didn’t need to be high, and the math closed.

what the theory predicts

Four claims you can check against the record rather than argue about:

1. A remake that holds the ludic dimension and only upgrades the presentation dimension outperforms its budget.
2. A remake that crosses the axis succeeds only where the original’s core verb was spatial.
3. Crossing the axis downward works for exactly the same reason in reverse, and almost nobody tries it.
4. Most “dead” genres are not dead. They are priced out. Drop the floor and they come back — which is what every 2D revival of the last ten years has been quietly demonstrating.

The remake business is the cleanest natural experiment this medium has ever run: same design, same audience, decades apart, with public unit counts at the end. Part two runs the theory against it.

Two dimensions was never a limitation the industry escaped. It was a medium the industry walked away from while it was still selling, because the accounting rewarded the bigger number and nobody had a word for what was being traded off.