It's actually not true that there's never been a hard fork. I know the Core FAQ says that, but it's incorrect. When we moved past Berkeley DB to LevelDB, there was an accidental hard fork that revealed limits in older nodes nobody knew about. But those limits had to go, and so the change was attempted again in a more controlled manner some time later. So it was known that the network would start producing blocks that older nodes would reject. That's a hard fork. It was scheduled in advance and went off smoothly.

The idea that hard forks are dangerous or irresponsible is a belief that is not well supported. However it's a rather good piece of Bitcoin Core propaganda to scare people away from doing what's necessary.