I used to always try for the best outcome but with this have it seems like half of the time a failure also leads to an amazing consequence and story.
Like this from act one in the Underdark:
spoiler
I had to find a hidden gnome that could supply me with gunpowder, but she was so much on edge that she lit up the barrel of gunpowder and blew up the whole room, leaving half of my party dead. A suicide gnome bomber. I couldn’t convince her that I was not an enemy. Reloaded just to see if I could successfully do it, but much preferred the first outcome of the dice roll, so had to reload and try 6 times until I failed again. What a game!
I only ever save scum pickpocketting. Years of DnD has taught me that fucking rolls up can often lead to more interesting developments. That being said, I also try to maximize success chances with guidance, finding advantage, building characters to be good at their jobs, etc
How DnD and PF and most RPGs does pickpocketing, it encourages save-scumming (or discourage use on a table game).
I think it would be better if the standard fail result wasnt “get spotted” but rather just “didnt get the item and made the target suspicious”. Then if you tried again and failed against a suspicious target, only then would you get spotted. That way the worst failure will only happen if the player push their luck. And pickpocketing/sleight of hand isnt the only skill that could have use of a more gradual fail state. If they push their luck on a lockpick check, it could jam the lock permanently. If they push their luck on a jump/athletics check, they’ll fall.
I’m no game designer, but I imagine that would encourage risky uses of skill checks (and in video games, not be so quick to reload just because of a failed check). It puts the risk management in the hands of the player instead of at the total mercy of a die throw.