three games to sell one
Entry three of three in “the dimension problem.” Part one builds the vocabulary; part two runs it against the remake business. This one spends all of it on a single case.
In 1997 Final Fantasy VII shipped on one console, in one box, for one price. It has sold more than 15.5 million copies, and roughly ten million of those were on the PlayStation alone.
The remake was announced in 2015, arrived in 2020, continued in 2024, and concludes on April 8, 2027 with Final Fantasy VII Revelation. Twelve years, three products, four platforms. Square Enix put Remake past seven million shipments by the time Rebirth launched, and 8.7 million including Intergrade as of March 2025. They have never published a lifetime unit figure for Rebirth.
Every piece of vocabulary from the first two parts is stressed by this one case. So let’s stress it.
the 1997 game was not a 3D game
This is the part almost everybody gets wrong, and the whole argument turns on it.
Final Fantasy VII is remembered as the game that took the series into 3D. Its presentation dimension did move — polygonal character models, rendered cutscenes, a camera that swooped through Midgar in the FMV. That is what everyone saw and it is what the marketing sold.
Its ludic dimension never moved at all.
The field screens are pre-rendered two-dimensional paintings with a fixed camera. You walk a model across a composed image. The image does not rotate, does not follow you, and cannot be interrogated — it is a screen in exactly the sense part one meant. Somebody placed every element in it, once, on purpose.
The battles are a menu. The ATB gauge fills, you read a state, you choose from a list. That is a compositional verb in its purest available form — closer to chess notation than to swordplay.
The world map is a diorama. It is abstract by design, a place-holder for travel rather than a place.
So Final Fantasy VII is the single cleanest example in the catalog of part one’s central distinction: modern presentation dimension, unmoved ludic dimension. It is the same trick Octopath Traveler and the Dragon Quest HD-2D remakes would rediscover as a deliberate strategy twenty-five years later. In 1997 it was the only thing the hardware could do, and it sold fifteen million copies.
what the remake actually changed
Put that against Remake and Rebirth and the move is not subtle.
Fixed-camera composed screens became a continuously navigable volume. The menu became a real-time action game. The abstract world map became an open world you traverse at ground level.
The core verb went from read the arrangement and choose to move, aim, dodge, position. Compositional to spatial. That is the axis crossing, and by the verb test in part one it is the predicted-fatal direction.
The tell is the ATB gauge. It survived. In the middle of a real-time action combat system there is still a bar that fills up and still gates your access to a menu, and pressing the button still slows time to a crawl so you can read it. They knew the menu was the product. They could not ship without it and could not ship only it, so the 1997 game is in there, running underneath, in a costume.
the director described the scale floor and didn’t name it
Why three games?
Tetsuya Nomura’s stated reason has been consistent since 2015: remaking Final Fantasy VII with modern technology is a massive undertaking, and compressing it into a single release would have meant cutting large amounts of the original — which would have made the remake pointless. Kitase floated two parts at one stage. They settled on three and confirmed it in 2022.
Read that again with part one’s vocabulary in hand. The scope of the 1997 game, rendered at 2020 fidelity in a navigable volume, costs more than any single product can carry. That is the scale floor, described precisely, by the director, in public, for eleven years running. He simply didn’t have a name for it, so it came out sounding like an artistic decision.
It isn’t one. The trilogy is an amortization schedule. Three purchase events, three marketing cycles, three revenue recognitions, because the axis made the thing too expensive to sell once. The narrative justification arrived afterward, as narrative justifications do.
the padding is the axis presenting its bill
Here is where the theory earns its keep, because it explains the one criticism that followed Rebirth everywhere despite a 92 on Metacritic — the best-reviewed entry in the trilogy and among the best-reviewed games of its year.
Reviewers who loved it still called it bloated. Too many minigames. Open-world activities that thin out. Padding between story beats.
Director Naoki Hamaguchi has estimated Rebirth contains roughly twice as much side content as main story content.
That ratio is not a failure of taste. It is arithmetic, and part one wrote the equation: composition becomes furnishing.
In 1997, the space between two story beats cost one painting. A single pre-rendered screen, and the player walked across it in nine seconds. In 2024 that same span of story is a navigable region — and a navigable region the player can cross in nine seconds reads as broken, unfinished, cheap. Volume demands to be filled. So it gets filled, with the only things that scale: activities, collectibles, minigames, side content at two-to-one against the story you came for.
Nobody at Square Enix decided Final Fantasy VII needed more minigames. The axis decided it, in 2015, and everything after was compliance. Every complaint about Rebirth’s pacing is a complaint about a bill that came due nine years before the game shipped.
the signature: acclaimed and underperforming at once
Remake scored 87. Rebirth scored 92. The better-reviewed game sold worse.
Producer Yoshinori Kitase said Rebirth’s PlayStation 5 launch sales fell below expectations. Square Enix president Takashi Kiryu, in the company’s financial disclosure, said profits in the HD Games segment did not meet expectations, naming Rebirth among them. And where Remake got a seven-million-shipment press release, Rebirth got no lifetime unit figure at all — which is itself a number, if you’ve watched publishers announce good ones.
That combination is the theory’s cleanest signature. Critics score craft. The market scores category fit. When a product is excellent by every professional measure and underperforms commercially, the usual diagnosis — “they should have made it better” — has nothing to bite on. You are looking at a category mismatch, and a category mismatch is invisible to a review score, because a reviewer evaluates the game in front of them rather than the one fifteen million people bought.
Watch what the publisher reached for. Not the category — the distribution. Rebirth went to PC in January 2025, then to Switch 2 and Xbox in June 2026. Revelation launches April 2027 on PlayStation 5, Switch 2, Xbox Series and PC simultaneously — the first mainline Final Fantasy to arrive on Xbox and a Nintendo platform at the same time since 1994.
More platforms is a thing you can buy. Category is a thing you’d have to admit.
the market-size problem
The comparison people reach for is “1997 was a different world.” It cuts the other way.
The original sold around ten million on PlayStation, against a lifetime install base of roughly 102 million consoles. Rebirth launched exclusive to PlayStation 5, which has now shipped more than 95 million units. Comparable bases. But the remake also had thirty additional years of brand equity, a total gaming market several times larger, a budget the 1997 team could not have imagined, and eventually three more platforms.
Remake and Intergrade together: 8.7 million.
The remake trilogy is not failing to find an audience. It found a different one, and it is smaller. That is not what failure looks like. It is what a category change looks like.
the objections, taken seriously
Three real ones.
“8.7 million is a hit by any normal standard.” Correct, and I’m not calling Remake a flop. The claim is narrower and harder: it is underperforming relative to the property it is built on and the money spent building it, and it is doing so in the specific pattern the theory predicts. A game can be both successful and a category error. Most of them are.
“You’re comparing one $60 game against three.” Yes — and that’s the amortization schedule working as designed. But a trilogy is a financing structure, not a value proposition, and it has a cost the spreadsheet doesn’t show: the audience has to re-decide three times. Attrition between Remake and Rebirth is exactly what the disclosures gesture at.
“Turn-based JRPGs just aren’t mainstream anymore.” This is the strongest objection and it’s the one that fails hardest on the evidence. The Octopath series is past seven million. The Dragon Quest HD-2D remakes are past four million combined, and Dragon Quest III was Japan’s only million-selling physical game of its year. Persona, Metaphor, Expedition 33. The compositional RPG is having its healthiest decade since the 1990s. It just isn’t where $200 million budgets go — which is the scale floor again, wearing a different hat. The category didn’t leave. The money did.
the game that wasn’t made
So what was the alternative?
Hold the ludic dimension. Composed scenes with fixed cameras, rendered at 2026 fidelity instead of 1997’s. Menu-driven ATB, unapologetically. The world map as a beautiful abstraction rather than terrain you walk. One game. One price. Out by 2021, at something like a fifth of the cost.
Would it have sold fifteen million? I don’t know, and anyone who says they do is guessing. But it would only have needed to sell about a fifth as many to clear the same bar — and it would have been selling to the people who bought it the first time, in the category they bought it in.
The most uncomfortable part: Square Enix knows how to build this. They have the pipeline, in-house, proven twice on Dragon Quest and once on Live A Live. The same company that rediscovered the exact trick Final Fantasy VII used in 1997, and made money with it, spent twelve years and three budgets walking the other way with its most valuable property.
Revelation ships in April 2027 and will probably be very good. Rebirth was very good. That stopped being the question somewhere around 2020.
What this series has been asking is what a product is for, and Final Fantasy VII has now answered it twice. In 1997 it answered with a menu, a painted screen, and fifteen million copies. The second answer took three games, twelve years and four platforms, and it is a beautifully made argument for something nobody asked to be argued.
You can execute at a very high level in the wrong category for a long time. The accounting is just slower than the reviews.