GPGPU is what displaced the Cell SPEs, and you wouldn’t count that as cores. In fact the PS3 originally didn’t include a GPU, the SPEs were supposed to handle it all, so in a way they were approaching the same solution from the opposite direction.

Was writing code for the PS3 genuinely difficult or was it also a skill issue? The counter-argument I've heard is that multi threaded programming just wasn't as common at the time due to minimal number of cores available on most consumer hardware.

Perhaps the PS3 would have fared better with a different generation of developers.

[1] https://youtu.be/dKYT6NzsUZQ?si=wzf5e5iBP3MEEJEn

First: One problem of the PS3 was that in a late stage of its development a (rather weak) GPU was "tacked on" (originally, the Cell was supposed to replace the GPU).

Now for your point:

Just have a look what scientists did on PS3 clusters using the SPUs of the Cell, and considering this imagine what creative things would have been possible for games on the PS3.

The problem was that at this time, game development got much less specific to one platform (in opposite to how it was for basically all the previous game console generations, though I am willing to admit that the original Xbox being very similar to a PC did give a hint concerning the direction the game industry was heading), so you wanted a much more "standardized" (typically meaning: similar to PC) architecture. The Xbox 360 was simply much more similar to PCs, so it was much easier to port engines from one platform to another.

> the people I've talked to who have worked for it said the architecture was a nightmare to work with.

For previous console generations this was less of a problem (basically every console was quirky). It is easy to find statements of developers on the internet how the PS2 was magnitudes more of a nightmare to develop for than the PS3.

So, before, being difficult to develop for a console was much more accepted in consideration of the unique features each console brought to the table. What did change was rather the expectation among developers to be able to easily port engines between platforms.

The nitpicking on language here is entirely due to the fact the distinction is no longer clear like it once was.

I think the real strategy was to make it harder for natively developed PS3 games to be ported to other platforms.

It was a sensible strategy except they effectively lost the console wars right out the gate by delaying their release a year. This alienated developers who didn't want to go through the effort of doing development the PS3 way when it was hard to program for AND had little traction among consumers.

If the PS3 had come out a year before the 360 things might have played out differently.

Sony's hubris around the PS3 was very risky. They caught up in sales by the end of that generation but early PS3 adopters unfortunately had to deal with fundamentally broken, unoptimized ports despite having superior (if over-engineered) hardware.

Also the PS3 disaster ultimately set the stage for the PS4 which had a drastically simpler architecture. PS4 was a lot easier to program for but if compared to a PC with equivalent specs it was overpriced and much more locked down.

This is exactly the problem Valve has solved. SteamOS presents a just works managed OS that is fully usable with a console controller. You hit the middle button on your game controller and the steam machine just turns on, wakes your TV, and switchs the input. The UX difference between Console and PC is shrinking to near zero.

Something that weighs on me is the fact that every tech device you own is constantly devaluing as it ages, eventually becoming worthless when support ends and the servers are shut down. It makes far more sense to own the least possible and get the most use out of each one. Unlike most things you own that only devalue as they are worn out and damaged.