In particular I'd add that:
>Shitty pay so actual devs just become software developers elsewhere and not gamedevs
Is exacerbated by a tendency to write games as individual projects where most or all of the team is contracted as temps, at worst entire studios are fired & hired every with every game as a regular practice.
And totally agreed on the rest, except I differ on:
>Monopoly of game engines and devs can no longer afford the cost/time to develop their own
That was IMHO both a bad and good thing. In the case of FPSs, for instance, licensing their engines unleashed a flood of games far more innovative than Unreal (Deus Ex, Undying, Rune, etc.) or Quake (Jedi Knight, FAKK2, Alice, etc.). Also, writing new game engines still isn't that hard, whether by rearranging licensed middleware to original effect, or fully custom, as compared to something truly implacable like writing moreso maintaining really a modern web browser engine.
>Diminishing returns on a GPUs performance and graphics
I think the bigger problem here is GPUs themselves. Fixed-function hardware acceleration made every game feel samey from VooDoo 1 to GeForce 3, nothing but triangles, bitmap textures, and Gouraud shading allowed. Shaders helped a bit, but haven't really been used for much.
Back in the software rendering era there was a somewhat more competitive field including other 3D techniques like voxels, fractals, quads, the aforementioned raytracing, etc., still in serious use. Even after 3D accelerators took hold, there was a last gasp of alternative hardware in the '00s that was just barely capable of doing legacy polygon rasterization in software as fast as a GPU (original PS3 design with double the SPEs & no GSX, Intel Larabee/MIC/Phi) for fallback in addition to more exotic graphics, but they fell ultimately through.
In theory, multithreaded CPUs with vector units, programmable GPUs with more flexible APIs like Vulkan & DX12, as well as APIs capable of bridging both like SPIR-V, make this fixable. But an entire generation of devs (and their toolchains!) have been brainrotted in ways that have yet to be undone.
>Don't care about graphics or technology (because in your mind these 2 things don't make better games, somehow)
There is an arguable point there, that putting more effort into one means putting less into the other, at least for small teams or risk-averse projects. A rebuttal I lean toward is that relaxed technical constraints allow the same results to be achieved with less effort.
Of course, I understood most of OPs argument was less about what underlying technological advancement there has been, and more about the inability to match even the level of realism in older titles.