a studio flexing at itself

the line that started this is from the director, yoshinori kitase, in the 1998 famitsu press cycle:

“these games take us 1.5-2 years to produce, so if we don’t set new challenges for ourselves, we can’t sustain our motivation through the development.”

that’s not “we wanted to make a great game.” that’s “we knew we’d get bored and would have to invent reasons to keep going.” read the whole interview record — kitase, ito, nomura, naora, across famitsu 1998, the gameinformer 2019 retrospective, the noisy-pixel coverage of the inside ff8 feature, the playstation blog 2021 piece — and a shape emerges. the shape is this:

every major ff8 design decision was a constraint, a provocation, or salvage from ff7’s reject pile.

i’ll defend that over the rest of this post. ff8 is the evidence; the principle is the point.

the reject pile

the antagonist — edea, the sorceress, the figure the entire plot revolves around — was a piece of ff7 concept art that survived in a filing cabinet between projects. kitase on the inside ff8 feature:

“there was actually a design for a sorceress that the character designer tetsuya nomura had created for final fantasy vii, and i really liked it. it ultimately wasn’t used for vii and was placed to the side, but it was very memorable.”

later, kitase asked for it back. that drawing became the basis for edea, and through her, the sorceress lineage that runs to ultimecia. the entire antagonist arc of the game is salvage from a previous project’s reject bin.

raijin and fujin are the same kind of salvage. nomura, in the same famitsu interview, on why they didn’t make it into ff7: “Raijin and Fujin were also characters I had created around FF7, and I had originally wanted to include them in that game. We thought they would overlap too much with the Turks though, so we scrapped the idea. Raijin and Fujin seem like characters that would appear in some old shonen manga, and that’s the image I had in my mind when I drew them.” the turks were the overlap. so raijin and fujin aren’t just unused ff7 characters — i think they’re tonally wrong ff7 characters. nomura drew them as old-shonen-manga archetypes, which is exactly the register ff8’s villains-in-uniform actually want. another 1998 file-cabinet decision, in plain sight.

the gunblade is a third shape of salvage — consensus, not inheritance. kitase, in the gameinformer retrospective, on nomura’s pitch:

“it was in his head somewhere — i have no idea where it came from. at first i just thought it would be a thing that could be both a blade and a gun. when he explained to me that actually, no, it was adding to the strength — when it comes down, it has the extra force of a gun to it — that made a lot more sense.”

kitase looks at a sword-with-a-gun, doesn’t get it, then gets it once nomura explains the gun isn’t a gun — it’s a force multiplier. that’s it. that’s the gunblade. it’s a consensus object — pitched by one person, almost rejected by another, talked into existence in a five-minute exchange. most of the iconic objects in shipping software are this. you’re not inventing them; you’re letting the room talk you into the bit you almost said no to.

ff8’s reject pile is unusually visible because the dev interview record is unusually honest about it. every team has a reject pile. kitase’s interviews go one further — they give you the moment of inventory: we kept this from a previous project, and now it fits.

the provocations

ff8 has at least three design decisions the developers knew would divide the audience and shipped anyway. this isn’t bitterness. it’s what kitase called “setting new challenges” so the team wouldn’t get bored. when you’re in year two of a 1.5-2 year cycle and your last game was a hit, you don’t make the same game again. you make a provocation.

the gil-as-salary system. in every final fantasy before ff8, you earned money by killing monsters. in ff8 you earn money — by being a student. you get a paycheque. it’s a stipend. kitase, in 2019:

“up until then, if you defeated a monster, you got money. but in final fantasy viii, we decided to give players salaries, so after a certain period of time passes, they get money. there were definitely people who weren’t happy about that, because it wasn’t what they expected or were prepared for it to be. but equally, i thought it would present a new appeal for people.”

note the construction. “people weren’t happy. i thought it would present a new appeal.” the salary was a provocation by design — and he says so now, twenty years later, with the satisfied tone of a man who’d do it again tomorrow.

the draw & junction system. in the same year, the battle designer of ff8 — hiroyuki ito, the person who invented the active time battle system — was also shipping a provocation. ff8 lets you draw magic from the environment and from monsters, then junction those spells to your stats. you don’t level up in the traditional sense. you stat-tune. ito, in the famitsu interview, on why:

“with rpgs, in the second half of the game it usually becomes trivial to defeat most enemies. with the draw system though, even late in the game, you can’t just bowl through everyone. if you want a particular monster’s magic, you’ve got to be more deliberate.”

and then:

“experiment on thousands of players, that’s another way of looking at it.”

(delivered in writing as a laughline. delivered here as evidence.) the level curve isn’t an accident. ito designed for late-game deliberation and called it out loud as a deliberate departure from jrpg convention. people who hit the wall in disc 2 and assume it’s broken are reading a provocation as a bug. (and — full disclosure — that discourse is one of the most fought-over parts of ff8’s reception thirty years later. kitase, on the record, reads forum posts titled “actually ff8 was pretty good” for the satisfaction of it. this is not a man who regrets the provocations.)

no voice acting. this one’s the most interesting because it’s the most consequential and least discussed. ff8 is a mainline final fantasy released in 1999 with no voice acting. every character is silent. every cutscene is silent. the emotional arc of squall and rinoa is conveyed through text alone, with characters mouthing things you can’t hear, against music that does things the game does not deserve. ff9 — the next mainline entry — shipped with voices the following year. ff8 makes that restraint a feature, not a deficit. kitase, in 2021, on what he would have changed:

“overall, i’m happy with the game, although there was one thing i was unable to achieve. during development, i wanted the characters to have voices. the visual and audio elements like sound effects and music were reaching film-like levels of quality, so the lack of character voices stood out as unnatural to me.”

read this carefully. this is a director saying out loud that the absence of voices — which forces you to read the text, feel the silence, watch squall mouth “whatever” — is the thing he couldn’t achieve. he wanted voices. he didn’t get them. the game’s most distinctive aesthetic quality is a shipping compromise that became identity. that’s not a provocation exactly; it’s a constraint that became the texture of the whole thing. the salary was a provocation. the junction system was a provocation. the no-voices aesthetic was constraint-as-identity — the empty space where a feature should have been, and what everyone ended up remembering about the game.

the constraints

constraints didn’t just shape ff8; several of ff8’s most distinctive decisions were the direct response to a single engineering problem.

hiroyuki ito’s junction system — the entire character-progression system of ff8 — is the cleanest example. ito, in the famitsu interview, on how the design originated:

“ff8 has significantly more realistic graphics than before. that realism meant that when you equip a weapon, it should show that individual weapon’s graphics on the character… but creating graphics for every different weapon would have been prohibitively difficult. i asked myself, well, if we can’t do that, what system could i have that would feel rewarding to players for getting a new weapon? and from that idea, i came up with the junction system.”

the entire magic-junction system exists because modelling sword #47 on squall’s back was going to blow the polygon budget on every weapon the player could equip across the entire game. instead of giving up on rewarding loot, ito redesigned character progression so a new weapon would feel rewarding by changing what stats you’d attach to it. the constraint didn’t stay a constraint — it became the game’s central mechanic and the entire discourse-defining design choice.

the ff7-to-ff8 tonal shift is the second. kitase, in the same famitsu interview, on the motivation:

“ff7’s story was a little gloomy, what with starting you right off in a slum, in a dark city without much sunlight. we all really wanted to do something brighter for this next game.”

this isn’t “the team chose pastel textures.” it’s a starting-state choice. ff8 starts in a garden, with a marching band, on a sunny morning. midgar, the slum, the dark city — kitase is naming a narrative-state, not a colour. the colour palette is downstream of the choice about the kind of world the player arrives in. ff8 feels bright because the world it depicts is bright.

the motion-capture-for-cutscenes decision is the third, and the one that most cleanly demonstrates kitase’s “set new challenges” line actually working. same famitsu interview:

“we added motion capture technology for the movies this time — if we had just used the same technology from ff7, i know the movie team would have gotten bored.”

a constraint — the movie team would have gotten bored — became a capability — ff8’s cutscenes have hand gestures and nuanced body language that ff7’s pre-capture animation couldn’t deliver. the constraint / provocation / salvage cycle isn’t just ff8’s pattern. it’s a method — when talented people are tired of the previous aesthetic, the only way to ship something interesting is to invert a constraint into a feature.

the fairytale substrate

this is the part of the story i didn’t know before reading the interviews and which i now think is the most honest thing in them.

kitase wanted ff8 to be wizard-of-oz underneath the sci-fi. from the 1998 famitsu:

“actually… and this is something i haven’t really said outside the company for the past 20 years, but i wanted the base of this to be in fantasy — in fairy tales. while visually, it presents as near future sci-fi, i wanted the base of the story and the world to be centred in fantasy, so i included sorceresses. in fact, i wanted to channel a bit of the essence of the wizard of oz, and that was what inspired me when i was creating the fairy-tale-esque, fantasy base of viii.”

this isn’t fan-press interpretation. this is the director, in 1998, telling you that the entire edifice of sorceresses, gardens, the dance, the parade scene, the moon-and-romance aesthetic, was his attempt to make a high-fairy-tale substrate under a near-future sci-fi surface. and then:

“ultimately, those elements became quite watered down.”

watered down, by his own admission. the wizard-of-oz pitch was not what shipped. what shipped is a sci-fi-with-fairy-tale-touches that works as a compromise. but the structural bones are still there — the sorceresses, the dance, the parade scene — and i think knowing what kitase was originally aiming at makes the cut version more legible, not less. you can see the lever-pulls he got to ship and the ones he didn’t.

the two scenes he obsessed over in that famitsu were the sorceress parade and the dance party. the dance party — the one that introduces the love interest — was nomura’s pitch for how squall and rinoa would meet. the most important scene in the love story is a dance the protagonist didn’t choose. that’s a deliberate choice, and it’s also a salvage from the wizard-of-oz pitch kitase says got watered down. the dance party is the bit that did make it through.

the cut that wasn’t, and the gag that should’ve been

two stories, side by side.

the dollet mission theme. the very first real combat mission of ff8 has a specific, recurring battle theme. the cutting room floor wiki records a widespread theory — i want to frame this as theory, not as fact, because that’s what it is — that the original demo version of the theme was cut because the new composition resembled the rock’s “hummel gets the rockets” too closely for square’s lawyers. the original demo track is still in the final game’s data, accessible only through a debug-room code, and has never been used. a legal-theory cut, not a design cut.

the fisherman’s horizon hand-slap. in FH, very early, there’s a moment where squall swings his hand at rinoa. it’s a stupid moment — a comedic gag where he’s being a jerk and she’s being irritating — and it lands as a character beat. the writer, kazushige nojima, told kitase at the time that it was a bad look for the protagonist to be physically aggressive toward the love interest. kitase agreed. they shipped it anyway. kitase, twenty years later:

“there’s this scene where squall and rinoa are talking at fisherman’s horizon, and i don’t remember what the conversation was about exactly, but rinoa says something kind of sassy to squall, and he throws his hand at her. she dodges, but even at the time, nojima-san was like, ‘he shouldn’t be hitting her. it’s really not good to have a guy hitting a girl.’ looking back, i wish i could change that.”

“this is the one thing i’d change.” in a game where lagoon’s cuts haunt the writer, the salary system remains a provocation, the wizard-of-oz pitch got watered down, and the gunblade was a literal pitch-meeting consensus — the one thing is a single gag from disc 2 that everyone involved already knew was wrong when it shipped. not the system nobody uses, not the cut content from lagoon, the gag.

the lagoon half-game

if the ff8 dev interviews have a single recurring ghost, it’s lagoon. he was supposed to be half the game — a parallel protagonist with equal screen time, equal plot, equal scene count. he became the past-tense flashback sequences.

kitase, in the gameinformer piece: “lagoon’s parts were much bigger. originally they were supposed to kind of be parallels of each other with the same amount of content.” nojima, the writer, in a later interview: “the staff worked really hard to make a map for lagoon’s parts too, but in the end it’s barely used in the final game. i always felt bad about that and couldn’t apologize enough to the staff.” nojima, asked if he’d write a lagoon spin-off: “he’d really need to be depicted as a clown for most of the game.”

lagoon is the example that breaks the categories. it’s not salvage from ff7 — it’s salvage from the current project’s own reject pile. constraint, provocation, and salvage — at once — for the same piece of content. that’s the whole shape of ff8 in one example.

so what’s the principle

i started with kitase’s line about “setting new challenges for ourselves so we don’t get bored.” that’s the real story of ff8. not “ff8 was great.” not “ff8 was bad.” ff8 was a studio at the peak of its confidence trying to design the next game while the constraints, the provocations, and the salvage pile kept intruding and refusing to be ignored. the system that emerged was deliberately off-key — junction instead of levels, salary instead of loot, no voices, the gunblade, the wizard-of-oz substrate that got watered down, lagoon shrunk to a cameo, gag that the writer flagged and shipped anyway.

every major design decision in ff8 reads, against the interview record, as one of three things:

1. a constraint that became a system. (junction, born from the weapon-polygon budget. motion-capture-for-cutscenes, born from a movie team that would have gotten bored otherwise.)
2. a provocation shipped on purpose. (salary. ito’s late-game deliberation. the ff7-to-ff8 tonal shift, which is constraint-as-provocation: tired-of-the-gloom-as-design-mandate.)
3. salvage from a previous project. (edea, drawn for ff7, saved for ff8. the gunblade, pitched and almost-rejected. the dance party, surviving the watering-down of the wizard-of-oz pitch.)

the categories bleed into each other — a salvage becomes a constraint becomes a provocation by way of how the team finally defended it. that’s the point. the categories are post-hoc analytical frames for a process where they were contemporaneous and tangled. the method is: when you have talented people, a tight timeline, and a previous game that went well, the constraint / provocation / salvage cycle is how you ship something interesting in year two of a 1.5-2-year cycle. ff8 wasn’t successful despite this cycle. it was interesting because of it.

— and if i’m wrong about this being the principle, the worst that happens is i’m wrong, and the writing was a good excuse to read the famitsu interviews. ✨

sources

1. kitase, nojima, ito, naora. 1998. final fantasy viii — 1998 developer interviews. famitsu (translated by clyde mandelin).link. — the only place you'll find ito's own voice on junction's design origin. also kitase on the *motivation* for ff8's provocations. this is the source that does most of the load-bearing work in the post.
2. kitase, yoshinori. 2019. back to school: the stories behind final fantasy viii. gameinformer.link. — the salary-as-provocation framing, the lagoon cut, the FH hand-slap scene. best retrospective for the *shape* of the development.
3. kitase, yoshinori. 2021. inside final fantasy viii. square enix developer feature (transcribed by noisy pixel).link. — kitase on the no-voices regret, the wizard-of-oz pitch, the 'tequila sunrise' colour palette. also the edea-was-ff7-concept-art story.
4. kitase, yoshinori. 2021. final fantasy viii: yoshinori kitase on taking the series in a bold new direction. blog.playstation.com.link. — the no-voices regret, in kitase's own framing. overlaps with the noisy-pixel piece.
5. the cutting room floor. 2026. final fantasy viii. tcrf.net.link. — the dollet theme / *the rock* theory, presented *as a theory* with the phrasing 'a widespread theory is that it was removed for legal reasons.' also the gerogero recolour and the selphie-nunchaku-to-'shinobu' localisation.
6. nojima, kazushige. (year varies). interview, recovered via gonintendo. gonintendo.com.link. — the writer still grieving the lagoon cut, with a clown-then-cool spin-off pitch. one of the only nojima-on-ff8 sources in english.