Traffic Management is a very complicated endeavour that we hand over to automated systems to route signals to – very much akin to light posts at a traffic stop.
But in the event of a power outage, if you watch people try to manage it themselves you’ll see it gets along “okay” until one person gets frustrated and decides they need to go a bit faster than what you are currently doing to manage the system. As they do this, other parts of the system (people) view this and start to do the same, creating points of contention and deadlock that did not previously exist because the system controlled the flow of traffic from one point to the other.
The same happens when we create hacks in our code to fix customer bugs which invariably end up “bypassing structure and rules” in order to get something done more quickly but maybe not as safe. All those customer hacks that go around things slowly add up until you have a system of implementations that are simply bypassing the issue.
In essence, you are then managing your traffic via hacks, which is not easy to manage at all.