Earlier quoted context omitted.
There is some middle ground though. I don't want my {critical infrastructure} to "fix things quickly and break things" at a critical moment, where I would have really needed that system. Those things can be scheduled and announced, so I can plan ahead and expect outages at that time.
> There is some middle ground though. Sure. But the problem is that there are incentives to keep pushing the middle ground closer towards eventual system collapse. It's never the right time to see if the backup generators can take the building load. But during real emergencies, it's amazing how common it is for the backup generators to not work for one reason or another.
How about a test outside normal working hours?
It is possible, to meassure the power output before and then plug in enough stuff, that draws roughly the same.
But yes, it is more convenient to not do it and continue buisness as usual and hope for the best.
My point is, that in most cases, you can test and fix critical stuff and also fix problems created by your fixes, if you make it an important issue and plan accordingly. However, I did not say it is necessarily easy.