Originally Posted by Phorbas

It's an unfortunate vicious cycle of not being able to release changes to the live environment regularly. In turn means each release includes a greater number of changes, which then has a higher likelihood of breaking something. When it does break, the answer is to add additional testing processes, which then means releases take longer to prepare - rinse and repeat.

Over time these processes get formalised, new structures established to perform each role, and then nobody has the whole picture nor the incentives to care when things aren't working.