Games SHOULD be a solved problem. There is no good reason for us to have to reinvent the wheel over and over. There are compsci white papers that neatly solve all of the big problems games run into.
But games are not a solved problem. There are multiple overlapping reasons why.
One is that gamedevs often just don't do the research. Why would they? The deadlines loom, the milestones have to be delivered, nobody cares if it's a hacky mess right now, surely management will give us time to fix it once they realize that it's broken--but if we don't deliver anything, the publisher cuts our funding.
This overlaps with scheduling and management issues. It turns out that writing good software takes up time[1], and the problem with games in particular is that they don't make money until they're released.
You don't write games like you write business software, where the other company paying for your milestones is the company that's going to use the software; that company usually has a revenue stream even without your software, so they don't have to care as much. For a game, though, there IS no revenue stream until the game launches. Every year that a game is in development without a release is a loss, and that pisses the board of directors right the fuck off, so that means the game needs to be out ASAP.
Because of this, games are often not given enough room in the development schedule to be made correctly. There's no time for research, testing, planning, or any of the other important parts of software engineering--we have to write this code NOW, or it doesn't ship. And if you read that source I linked in the footnote, you'll know that this produces rotten software.
This is compounded by the kind of one-upsmanship that is created by such an environment--leading to a phrase I've heard from friends in other companies: "Very optimistic people who are no longer with the company made this decision". You get into a situation where people made promises to impress the publisher, claiming that they can turn out a game in an impossible timeframe, and that got them fired--but now you're stuck cleaning up the mess, and the publisher has already wasted a lot of money on the years spent thinking it wasn't a mess.
Mix in the siloing of information (because all of this shit is proprietary) and, despite all of the problems being solved on paper, nobody's solved video games.
[1]: https://jacobian.org/2022/sep/9/quality-is-systemic/