It’s a sliding level of personal involvement and drive across teams and across features. Of course every game is always over-scoped, so production is a matter of figuring out how to keep everyone balanced with work while yelling at designers to stop adding features and then murdering their babies. A specific lead or really, any specific person should be capable of driving a feature they really care about, but if they slip or lose attention just one weak link can wreck it – and good luck clawing back some other team’s time to redo a deliverable that was 80% quality.
In this case, for the haiku feature to work I think a designer and writer would need to be the main drivers. If a designer owned it, they could wrangle the art production but may have ended up writing these (terrible) haiku themselves. Or if they left a writer to it, the writer could have been left with a problem: make modular haiku (dependent on contextual information about the area around the interactable that may not have been finalized until late in the project!) without the time to fix what is actually a tricky problem: modular poetry. It’s a lot easier to read these and realize they’re terrible then to write 2 dozen modular haiku! If it was writer-led, then they would have spent tens of hours chasing down world design and art production to make sure the feature was being supported correctly and that’s not usually in the core competency of writers, who tend to be kind of silo’d and isolated.
And it’s a case where the non-personality of Jin didn’t help them. A character like Geralt has a simple, recognizable personality that writers can queue off of. Instead of ‘write an idealized modular haiku’, the problem would be, ‘write a modular set of quips Geralt might make’, and we’d have a framework for understanding dumb outputs as Geralt making a joke or just giving up. With an idealized blank canvas, the haiku can’t fail and reveal something about Jin; it has to stand on its own right.
I think team organization can explain a lot about the games put out, actually. For example, Ubisoft has the most distributed, factory-like game production model in the world. More than anyone they’ve managed to make worldwide studios cooperate in shipping enormous games and they’ve been able to reuse tech and features across games (like their great water sim). The ‘creative council’ is a small enough, tight enough team that they generally have more interesting game concepts than other AAA studios – tightly-bound, specific looks at historical periods that should theoretically interest me. But the distributed, modular production makes it almost impossible for any one part of the game to rise above mediocre, and their games succeed only in scale and never ever have any specific genius inside them.
A team like Bethesda (post-Oblivion) is driven by the world and dungeon designers, who are responsible for quest design (and writing! (!)) and they structure features off of the world events and quests the team of designers are building. Because of this, they can build much more content at their scale because designers are left alone to tinker with levels and quests. But they’re bad at maintaining a consistent tone or theming or features that drive the game because they’re focused on small contained content objects. This production is probably the same as produced older, more directed games like Daggerfall and Morrowind but when the team moves from a dozen designers covering world, quests, dungeons, system to three or four-dozen the ability of a strong personality like Michael Kirkbride to influence the entire game is muted.