Isn't there a catch-22 here? The system is a republic which writes its own requirements; if it's been compromised and exploited by outside systems to the point its kernel level processes are funadmentally corrupt/compromised, how can you trust it to autocorrect in a way that actually fixes the fundamental issues? The tools designed for "the people" to use and interact and influence the system have been neuteralized with mathematical precision (assuming a point where they worked as specified, there is an argument to be made that the system has always given lip service to the end user while really only catering to management).

I think you only have a chance at implementing change if you're able to locate and leverage an exploit that the viruses within have either overlooked or can't completely secure against.