Honestly when was the last time someone created a "good" AAA god game anyways? Simulation games are only really decent if they're made by some really obsessed indie guy.
The real issue at the heart of this is that there's "experimental" design, where you use untested ideas, and it can come up with something really cool, but it's more likely to end in colossal failure as you find out whatever you did doesn't actually work. When it's just one guy building a simulation, that's low-risk: you're only risking the effort of that one guy, but when there's
also $80 million dollars invested in artwork, models and marketing, then the "experimental" part of the design becomes a huge financial risk.
Molyneux's problem is that he approaches things as if he's still that lone 80's dev building Populous for 8-bit machines, where you can cheaply try out new ideas. And back then graphics and stuff were so low that you could specify huge ideas and they were represented so crudely that people would accept it. Now, with hi-res textures and movie-quality 3D, idea execution is a lot more expensive. Molyneux is leading large projects now, and the people on the project and the financiers tell him "it just can't be done". And they're right, so the grand ideas get whittled down to achievable, measurable results.
In Spore terms, it was asked why the evolution model is mainly cosmetic. Because we
know how to move a thing around that's just a physics blob that has a cosmetic skin attached. The code to do so is tested, and it runs fast. We
don't know how to build an entire physics system around evolving, articulate beings, and the frame rate would end up being abysmal. Sure, you can try, but where is one single working example? The
odds of making it work are very low. And you have a budget in the 10's of millions that will go down the tubes if your magic experimental ragdoll alien physics system doesn't work, and you won't be asked to make any more games after pulling that shit, that would be your last one. This is why AAA devs with big ideas end up making fairly
safe things where they're actually constructed out of known, working elements: you have to prove that every system you're going to build
will work before production begins.