I read many posts saying that devs should care about save scumming and avoid it at all costs.
I really don't understand and reading articles about it doesn't really help.
According to me save scumming is just a way of "cheating" to write the story we want.
I'm not the biggest save scummer (?) but sometimes I'm doing it. In exemple when playing games like XCOM or The Darkest Dungeon because I don't want my high level character dead.
I also do it sometimes in RPGs when the event takes a path I don't like.
In BG1/2 I also do it sometimes if one character is perma dead, or if 5 characters out of 6 die in battle.
I just reload and do it again.
In Total War I'm ALWAYS saving before any battle and I reload the auto save before each turns A LOT.
I guess that's what's call save scumming.
I could also do it in BG3 in exemple if I miss a roll that leads to a combat when I was planning a pacific solution.
As I understand it, save-scumming isn't about reloading to try to do better or exploring different options - save scumming is about abusing RNG to cheat your way through the content, and not engaging with design in intended way. So less about reloading battle until you do well enough, and more about spamming finger of death that you have no chance of landing until you roll 20. Not playing iron-man is not the same as save-scumming. Bypassing mechanics through abuse of RNG: is.
I think people who will complaint about save-scumming are people who are susceptible to it, and don't enjoy it. How game design influences players is something devs should very much consider, though of course it is impossible to make a tailor-game made game for everyone. If we talk about skill-checks things few things to consider would be:
How many people reload if they fail a check?
Is failing a check our character is highly proficient with in some way engaging? Is winning a check we have little chance to succeed in rewarding?
Does RNG in skill checks make playthroughs more varied, or do players gravitate toward the same path, as their character choices are being bypassed through save-scumming?
Personally, I had a fine experience with BG3 skillchecks - most of them went as expected, there were few unfortunate failures, and few unexpected wins. With straight 20 roll, it is a matter of luck, and I can see things going terribly wrong for someone (assuming it is a flat 20d roll - FiraXCOM1&2 cheats quite a bit with their percentage, and players still complain about those few misses - something that I didn't find that commong in my many, many playthroughs. Some people will just dislike RNG). More importantly, I don't recall BG3 failing me into an instant failure state - having things go into unexpected direction is A-OK in my book, killing me (I mean pernamently - game over/try again) for rolling a wrong number is not.