I tried to read this, but this post was so obviously written by AI, it made it hard to proceed. There's a few human written paragraphs written in the first person that have a jarringly different flow form the AI written portions.
Art costs are probably the most predictable part of game development (they’re also a large part of it). If you go to an outsource studio and tell them what you want you’ll get a variety of quotes depending on the level of quality you want.
The articles point us mostly right but this isn’t new to the AI age.
> None of them could tell me the cost of a decision that got reversed six months later.
Absolutely everyone in a game studio knows that the cost of a bad decision scales with the number of people working in the project, and all the good leads know the decisions that are easy to revisit. There’s only two terminal decisions in a game project (IME - I’d love to know if there’s another). Not finding the fun, and scaling too early.
> Headcount grew from roughly 100 people in 2015 to around 400 by 2019
Skull and bones shipped in 2024. Pandemic notwithstanding, this checks the “scaled too early” box
If making game assets got really cheap (because of AI or whatever), the budget would just be spent on marketing. In theory you can spend infinite amount of money on marketing.
The "a Game" in the title is unfortunate. The author seems to have only heard of AAA games (or in the case of Skull and Bones, AAAAA), but that entire industry is becoming less relevant by the day. Many of the top earners come out of the indie to AA cohort now.
At that level, using AI is extremely tempting because we often entirely lack the capacity for whatever it can slop out. Personally, I can resist the temptation because I do employ a human artist, and even if he's not great at some particular output we can still produce something good enough.
But many, many more games never came into being at all because there's just a programmer, or just a designer, or maybe a team of 5 with just one artist. Not existing is certainly a failure, no?
So, now we're seeing a flood of terrible slop from solos churning out their first game. All the noise sucks for everyone who has some expertise and an existing team. I think it's going to suck a lot worse for all of us -- indie, AA and AAA -- in a year or two when some of those slop producers actually become good.
This part seems really true:
"If execution becomes genuinely cheap, then being wrong becomes cheap, and a studio can afford to explore more directions before committing to one. Run six prototypes instead of two. Kill five of them in week four rather than month fourteen."
This really is huge. My own plan for the future is to make multiple vibe coded/art prototypes and then redo the chosen one from the ground up.
Yeah, plenty of game projects fail but just not publicly.
My game projects failed before I could start them because I need to work to pay rent and don't have time to spend years tweaking graphics of indie games.
Claude is now so good I actually turn short stories I write into games for myself.
There’s a great thesis that can be analogued against the speed of vibecoding here, but in terms of the thesis in the title, there is a selection bias, there’s millions of games that failed because the artist was so slow, they never produced the necessary deliverables, it’s just that indeed you never see them.
never delivering is just an extreme case of slowness.
I can think of quite a few games with extended development cycles where they scrapped what I originally wanted and eventually released a game that, while still decent, "underperformed" because it was essentially tasked with turning the profit of two or three games instead of one.
Gosh, just finish the first game and ship it. Then do the same rework you were going to do and ship that too. I will happily enjoy both cakes.
It's crazy to think that the leadership of multimillion-dollar publishers and developers operate with the same conviction and decisiveness as an unfocused art student.
EDIT: I look at something like Sony third parties during the PS2/PS3 transition vs Nintendo around Y2K. SO many PS2 games got delayed and ported to PS3, getting roughly the same sales on higher expenses. Square Enix even infamously did it with the PS3/PS4 transition, scrapping 7 years of work on Final Fantasy Versus XIII to turn out XV in the following 3 (though, in a sense, they learned their lesson, since XV was basically released half-completed, the Royal edition finally adding in most of the planned features).
Compare to Nintendo releasing several of their N64 games in Japan with missing features that would be added in for a subsequent US or GameCube release (most notably Pokemon Stadium and Animal Crossing). Instead of trying to desperately claw back extended development expenses, they double-dipped.
I tried to read this, but this post was so obviously written by AI, it made it hard to proceed. There's a few human written paragraphs written in the first person that have a jarringly different flow form the AI written portions.
[flagged]
Core topic seems worth discussing but I gave up halfway thru after reading too many LLM-isms.
Art costs are probably the most predictable part of game development (they’re also a large part of it). If you go to an outsource studio and tell them what you want you’ll get a variety of quotes depending on the level of quality you want.
The articles point us mostly right but this isn’t new to the AI age.
> None of them could tell me the cost of a decision that got reversed six months later.
Absolutely everyone in a game studio knows that the cost of a bad decision scales with the number of people working in the project, and all the good leads know the decisions that are easy to revisit. There’s only two terminal decisions in a game project (IME - I’d love to know if there’s another). Not finding the fun, and scaling too early.
> Headcount grew from roughly 100 people in 2015 to around 400 by 2019
Skull and bones shipped in 2024. Pandemic notwithstanding, this checks the “scaled too early” box
If making game assets got really cheap (because of AI or whatever), the budget would just be spent on marketing. In theory you can spend infinite amount of money on marketing.
The "a Game" in the title is unfortunate. The author seems to have only heard of AAA games (or in the case of Skull and Bones, AAAAA), but that entire industry is becoming less relevant by the day. Many of the top earners come out of the indie to AA cohort now.
At that level, using AI is extremely tempting because we often entirely lack the capacity for whatever it can slop out. Personally, I can resist the temptation because I do employ a human artist, and even if he's not great at some particular output we can still produce something good enough.
But many, many more games never came into being at all because there's just a programmer, or just a designer, or maybe a team of 5 with just one artist. Not existing is certainly a failure, no?
So, now we're seeing a flood of terrible slop from solos churning out their first game. All the noise sucks for everyone who has some expertise and an existing team. I think it's going to suck a lot worse for all of us -- indie, AA and AAA -- in a year or two when some of those slop producers actually become good.
This part seems really true: "If execution becomes genuinely cheap, then being wrong becomes cheap, and a studio can afford to explore more directions before committing to one. Run six prototypes instead of two. Kill five of them in week four rather than month fourteen."
This really is huge. My own plan for the future is to make multiple vibe coded/art prototypes and then redo the chosen one from the ground up.
Plenty of games have failed for lack of content.
Sorry didn't read it, because the title screams "survivorship bias".
Yeah, plenty of game projects fail but just not publicly.
My game projects failed before I could start them because I need to work to pay rent and don't have time to spend years tweaking graphics of indie games.
Claude is now so good I actually turn short stories I write into games for myself.
Holy what kind of writing is this, obviously there's a lot of stuff written by a human but then roughly interleaved is clear LLM garbage.
There’s a great thesis that can be analogued against the speed of vibecoding here, but in terms of the thesis in the title, there is a selection bias, there’s millions of games that failed because the artist was so slow, they never produced the necessary deliverables, it’s just that indeed you never see them.
never delivering is just an extreme case of slowness.
I can think of quite a few games with extended development cycles where they scrapped what I originally wanted and eventually released a game that, while still decent, "underperformed" because it was essentially tasked with turning the profit of two or three games instead of one.
Gosh, just finish the first game and ship it. Then do the same rework you were going to do and ship that too. I will happily enjoy both cakes.
It's crazy to think that the leadership of multimillion-dollar publishers and developers operate with the same conviction and decisiveness as an unfocused art student.
EDIT: I look at something like Sony third parties during the PS2/PS3 transition vs Nintendo around Y2K. SO many PS2 games got delayed and ported to PS3, getting roughly the same sales on higher expenses. Square Enix even infamously did it with the PS3/PS4 transition, scrapping 7 years of work on Final Fantasy Versus XIII to turn out XV in the following 3 (though, in a sense, they learned their lesson, since XV was basically released half-completed, the Royal edition finally adding in most of the planned features).
Compare to Nintendo releasing several of their N64 games in Japan with missing features that would be added in for a subsequent US or GameCube release (most notably Pokemon Stadium and Animal Crossing). Instead of trying to desperately claw back extended development expenses, they double-dipped.
You can't "fail" if you don't ship. /s