If success is actually impossible, then you wouldn't roll at all.
But you're doing the rather typical thing of openly taking the least-charitable possible interpretation and presuming it must be true to skewer it, rather than asking, "Wait, does that mean success is guaranteed?"
The poster to whom I was replying specifically said (paraphrased from memory) if there's no consequences for failure, just give them the success rather than roll for it.
And yet sometimes the true consequence of failure is simply that you didn't succeed - it's the success that has consequences, and you didn't achieve it. You didn't find the main treasure hoard. You didn't find the hidden slave pens. You didn't climb the cliff and now have to go 40 miles out of your way to get to the top.
And that is a pretty boring consequence. I find it rather frustrating to see people hold it up as though it were some awesome achievement of gameplay that you made people jump through hoops in order for literally nothing to actually happen.
They jumped through those hoops in the hopes that something would happen. It wasn't guaranteed.
No, it isn't. On both counts.
Specifically, you are conflating two different kinds of "frustrating." On the one hand, there is, "I, the character, am trying to succeed, and not reaching success". That, I agree, is the point of rolls. But it's not the point of rolls where the only results are "things proceed without issue" or "nothing happens and we just spent the past 2-5 minutes literally not seeing anything happen", which is frustrating from the, "I, the player, have just wasted several minutes of time literally achieving nothing whatsoever, not bad, not good, nothing."
The former type of frustration is a good thing, and should happen with a reasonable frequency. (Different people, obviously, disagree on "reasonable" frequency. I'm pretty confident your threshold is much, MUCH lower than mine, for example. But the idea that there is a threshold isn't in question between us.) The latter is bad and should be avoided as much as possible.
If you're inhabiting your character as its player then the two types of frustration - in-fiction and at-table - should align very closely, thus making it difficult-to-impossible to have one without the other.
I'm fairly confident that is not true, because there are at least three other options. One, no chance of failure, unless you force characters to roll to check to make sure they can walk across rooms, open doors, shave without decapitating themselves, etc. Two, where neither success nor failure actually matters in any way whatsoever. And three, where the action in question isn't a matter of success or failure, but rather degree of success only (this is rare, but consider for example magic missile, which, AIUI, explicitly doesn't use an attack roll nor a saving throw in any edition. That's something where you ask for a roll, the attack just hits...or just misses, if the target throws up shield in time.)
1 - in all cases here I'm assuming the task or intent in question is something where neither success nor failure is guaranteed.
2 - whether or not something matters often isn't evident until well after the fact; in the moment, we have to assume it all matters
3 - degree of success is tangential; for me the roll can inform degree of success in some situations, while others are pretty much a binary pass-fail.
I mean, in my experience, it holds up extremely well, on both ends of the table.
Failure is still failure, it just means SOMETHING happens. The world continues spinning. Maybe you """succeed""" in a way that is completely hollow, like "you found the secret entrance eventually, but by the time you were done, the cultists were LONG gone and knew not to leave evidence behind because they could hear you the entire time." Maybe you fail, and now that's created a problem: "In trying to disarm the trap, you've not only set it off, you seem to have triggered some kind of deeper, more magical defense system. That's...really really not good. You can hear strange noises in the distance. That's probably worse."
And if the fiction is set up that way with layers of defenses, alarms, etc., that's great.
Problem is, it's not always set up that way. Sometimes the only consequence of failing to disarm a trap is "nothing happens" other than the trap remains in place as a potential threat.
If failure would contribute literally nothing whatsoever to the experience of play other than delaying the party's next effort, what is the point of rolling? Like seriously. If literally nothing comes of failure--not even expending resources, genuinely actually nothing happens--why should you roll? Save the rolling for when it's actually interesting to fail, and interesting to succeed. Rolling to avoid trivial stupid failure like "you walked across the room wrong and flung yourself at the floor" isn't helpful, and there are a LOT of things I've seen GMs ask for rolls about that really should not have required anything of the kind.
Let's take a trapped lock as an example, DC 12. There's a perhaps-simplified range of possible outcomes on an attempt to disarm the trap:
1 - trap disarmed no problem (14-20)
2 - trap disarmed but doing so made the lock unusable (success with complication e.g. the acid that was supposed to melt the thief's fingers melted the lock instead) (12-13)
3 - trap not disarmed but is still there as a threat (i.e. nothing happens) (5-11)
4 - trap is set off (thus requiring possible resource/spell use to fix the consequences) (1-4)
A true-binary roll here would only have 1 and 3 as possible outcomes, or maybe 1 and 4. The way I'd rather do it, though, includes the whole range of outcomes and gets there by only one roll (plus maybe a saving throw if needed on the '4' outcome).
Note however that the fail-success break point remains at 11-12 - success with complication still needs that root 'success' roll in order to occur.
