But it has stakes beyond simply the picking of the lock. There are other potential consequences.
Which points out what might be a difference here.
When I think of what happens in the fiction due to a roll, I'm thinking of the immediate stakes on which the roll is made. Do I climb the cliff or fall. Can I open the lock or am I stuck here. Can I sneak past the guards or do they notice me. Those are stakes.
Anything and everything beyond that falls under downstream consequences, which are not the same thing as immediate stakes and are also not always (as in very rarely) 100% predictable or guaranteed. They can very often be changed, mitigated, or even eliminated by subsequent actions provided those actions are themselves successful.
Falling due to a failed climb doesn't necessarily even mean I'm going to go splat at the bottom - I might grab a handhold and slow or stop my fall, I might have someone below willing to take the risk of trying to catch me, someone (including myself) might have a magical means of preventing or easing my fall, and so on; and despite the fairly obvious train of causality in the fiction these would all - if available - be declared and resolved as independent actions.
In hindsight, it might come down to did I go splat because I fell or did I go splat because you didn't catch me.
Not at all... I showed how it could have a consequence other than just an unopened pickle jar.
A downstream consequence, not related to the immediate stakes of opening the jar.
But as far as games go, I wouldn't expect a game to even worry about such mundanity, or if it did come up for some reason, I wouldn't bother gating the opening of the jar with a roll. I'd just say "You open the jar... now what?"
Ditto re worrying about the mindanity in an actual game. The example is intentionally very mundane in order to remove all the baggage surrounding the break-in example we've been using.
Mitigating risk is one thing. Refusing to proceed unless all risk can be mitigated? That's another. You've described plenty of instances of play where players have been less cautious. You've even advocated for PvP with "Let them fight, says I" even though this would obviously be against their best interests.
If your game actually grinds to a halt once danger rears its head then I'd say it's probably the worst D&D game ever. I don't expect that's the case, though, so why you're trying to argue this is beyond me.
But this just displays a continued misunderstanding of the idea. If we took a session I run of Stonetop and one you run of your version of D&D, then yes, a roll in Stonetop is likely weightier than one in D&D. But there will almost certainly be fewer rolls.
If the fiction suggests little to no chance of dangerous consequences, then I'm not going to ask for a roll. I'm only going to call for a roll in Stonetop when there's something at risk. If there's no risk at all, then I'm simply going to allow the character to succeed.
Also, the idea that if a character fails, they're getting "hosed" is just... I don't know... it seems adversarial even for D&D.
A lot of the examples we've been given of fail-forward in this thread seem to be doing just that: hosing the characters by adding in extra complications rather than just narrating a straight nothing-happens result. Some of the games even tell the GM do this, put in terms of "always drive them toward conflict" or similarm which makes it seem like a very stressful way to play.
It's a consequence has been my point. You guys are fighting tooth and nail to avoid saying what's incredibly obvious.
Consequence, not stake; and only visible as such in hindsight after any other possible consequences or outcomes have been somehow eliminated.
So Thief A is in lock-picking class with his guildmaster. They're practicing picking a lock.
Thief B is in a dark alley about 20 feet from a well-lit street. It's late at night, but there may still be people about, certainly the watch is still on patrol. There's a light rain, but it's only a little more than a drizzle. Thief B is trying to pick the lock of a merchant's office so he can steal some documents for a rival merchant. He'll also see if there's anything of value that can be taken.
According to you, these situations are identical?
No, in that the surrounding conditions will be changing the odds of success.
Yes in that the actual task resolution still comes down to a binary succeed-fail on whether the lock is picked or not; and also in that the immediate stakes - whether or not the lock is opened - are the same.