> Systems are as decentralized as the people running it;
No, this is the worst case scenario, the core developers of a technology like Bitcoin have the power to, arbitrarily, govern the project and change the code or protocol without the people noticing or caring about it. There are obviously limits that are not acceptable to the people (e.g. SegWit2x and Bitcoin Cash). The governance of a protocol is mostly centralized. Perfectly decentralized systems are decentralized systems that never change. If they change it is because there is governance that plays in the power context.
Why would there be no 'people noticing or caring' about a change?
Everything I see, from Canaries to codesigning, to simple git mirrors make it easy to detect such fraud.
Also, there are many unofficial Bitcoin implementations. And with 'core' being the only 'spec', I know from personal experience that a lot of devs of those alternative implementations, make a lot of 'eyeballs'.
> Why would there be no 'people noticing or caring' about a change?
Because you cannot assume that node runners and miners are spending time in reviewing commits or entering in discussions except when there are issues that are highlighted and amplified by media such as SegWit2x.
There is just no such thing as software that can't change; I realize this might mean decentralized systems are simply impossible, but "it is what it is"?
I do feel you just don't understand this still, though: if the miners all upgrade and you don't it isn't like they can force anything down your throats as you didn't upgrade.
I feel like your mental model is somehow "if 51% of everyone does something invalid then the system is now invalid", but what that actually means is that those people are rejected from your consensus.
So, if all the miners decide to run a different software this is effectively just a different distributed system. Your computer, which is checking everything, rejects all of them and keeps playing with the people who are playing the same game.
You aren't trusting anyone or assuming anything: that wouldn't be decentralized at all really. Now, is any of this efficient? No. But that wasn't a goal ;P.
No, this is the worst case scenario, the core developers of a technology like Bitcoin have the power to, arbitrarily, govern the project and change the code or protocol without the people noticing or caring about it. There are obviously limits that are not acceptable to the people (e.g. SegWit2x and Bitcoin Cash). The governance of a protocol is mostly centralized. Perfectly decentralized systems are decentralized systems that never change. If they change it is because there is governance that plays in the power context.