The feeling that soft forks are more desirable/better than hard forks is not as simple as it seems.
With the drama of BIP444 I've seen a lot of people confused about the nature of soft-fork vs hard-fork. This might help:Back in the block size war we had a lot of similar confusion, mostly because it isn't really very clear cut. Some people talked about "evil soft forks" to try to get rid of the simplistic notion: "soft forks are much better and hard forks are much worse" which tends to persist (naturally). The problem is that whether a fork is contentious or not is much more important than whether it fits the "soft" or "hard" technical definition.When a fork isn't contentious, then the soft vs hard distinction (restricting the ruleset or relaxing it) really matters a lot, because "passive" network participants (imagine a piece of hardware that will never get updated to a new bitcoin version, in the extreme) will be fine with the first and not with the second.When a fork is contentious, and miners, following economic incentive, end up choosing different rulesets to support different bitcoin users, then the chain genuinely forks into two histories. The fact that one ruleset is more restrictive than the other is part of the story but doesn't change the fundamental point that we have a chain split. This of course did actually happen meaningfully with bitcoin-cash in 2017. The term "evil soft fork" was some people trying to shake others of the misconception that if a fork is "soft" it's not coercive and not forcing action on anyone; that's definitely not true if the fork is contentious.