Exactly. Why not make them crit? It's going to be up to the DM anyway to define what a "critical success" means on a skill check. There's no hard rule like the extra damage that comes with crit successes on attacks. The DM gets to choose what a critical success on a skill check actually produces. The DM can easily just make sure the crit success isn't game breaking.

Your players are in an audience with the king. The bard tries to be funny and tries to convince the king to give him his crown and hand the kingdom over to him. Actually making the bard the new king would break the game. But maybe a critical fail means the bard gets sent to the dungeon to be tortured for daring to make such a request. A critical success means the king will grant the bard one "wish," ie, any reasonable single reasonable request that is within the king's power.

The whole situation is fully in the DM's power.

D&D today is almost an unrecognizable game from its first incarnation in the 70's, though. I'm not really seeing the parallels to war games other than the fact that you have the option of using a battle map in combat, which is hardly unique to D&D.

To borrow your analogy, no one drives the Model T today, but cars still have 4 tires and a steering wheel.

Yeah, the people that do rules as written, or follow a book for a campaign to the letter, to hard often end up taking the fun out of it.

My first ever campaign I was an outlander ranger with high survival. We started in a swamp and it was written "pass survival check, if fail, roll to go in a random direction". I somehow failed 7/8 rolls with +7 (bad luck). We spent the whole session going round in circles and ended up further away from our objective than we started.

I felt awkward/stressed, and the others just felt bored/frustrated.

Chatting with more seasoned players afterwards they were like "yeah, that shouldn't be how it normally goes, but it's not your fault, DM should have a fail safe for stuff like that. First rule is 'is it fun'. Just cause the campaign says 'do x' doesn't necessarily mean you should if it's not fun for anyone"

Not to mention which game you're actually playing.

Some players don't ask.

Once in a blue moon, an impossible check can impress a scale of difficulty on the players.

D&D example: a player with a high bonus attempts an Arcana check to figure out an enchantment and rolls well, up to a natural 20. I let the players have their moment of joy. Then I make a big deal of telling them they don’t have any idea what’s up with this enchantment. I really talk up how weird/complicated/confusing/impenetrable the enchantment is.

I’d be trying to prompt emotions I want the players and PC to share. Frustration, inadequacy. The players would viscerally know they need to try a different approach.

And because I gave the check a decent chunk of game time, it has more narrative weight. An interactive skill check is more substantial in the player’s mind than a monologue on the task being impossible, particularly if it stands out because they fail that check despite a super high result.

It’s a niche scenario, I admit. Most of the time just don’t ask for the check.

In addition to what the others have said, I think degrees of failure are often a fun thing to introduce whether they are in the rules or not (I'll assume D&D 5E). It might be that a 20 with your +3 athletics isn't enough to completely leap over that huge gap, but you manage to grab a handhold a few metres below the edge. You'll have to take a turn or two to climb up, but you're okay. The cleric's roll of 3 with a -1 athletics, on the other hand, sees him plummeting to the bottom and taking a heap of fall damage

Maintaining the illusion? Not revealing the (impossible) DC?

Because I don't have everyone's modifier for every skill, ability, saving throw, and attack memorized off the top of my head, nor do I have magical foresight into whether or not they will choose to use abilities that would add more additional points on top of those modifiers.

Mutants and Masterminds has (effectively) a +5 if you roll a 20, but no extra penalty for rolling a 1.

Agreed, auto success on a skill check nerfs challenges.

If the DC is so high that the PC doesn't succeed with a 20, it seems too random to give it to them.

Then again, it depends on the situation: a nat 20 trying to convince the penny pinching tavern owner to give you a discount seems like fun even if the DC should be infinite; but when dealing with something story related, I'd stick a little closer to the rules.

I 90% agree. I think most of the opposition to this comes from people exhausted with habitual boundary-pushers who think that a nat 20 means they can get away with defying the laws of reality.

Like, no, a nat20 persuasion does not convince the merchant to give you half his stock and all the money in the register... He would go broke and he's got a family to support, along with his own survival that your nat20 does not also convince him to stop caring about.

But at the end of the day, a lot of GMs who are sick of that need to be sent the dictionary page for the word "no." The occasional use of it really does improve the quality of the game, and I'm sure plenty of players will appreciate not letting aforementioned boundry pushers continue to waste time on impossible pursuits that do nothing to move the game forward.

D&D is that way, though. Every time you see a natural 20 for anything that isn't an attack does not automatically succeed unless people are using homegrown, which they often are.

Hello fellow Pathfinder!

Dating back to 3rd critical skill checks in D&D suck because a lot of skills are written as pass/fail.

Example: picking a lock. If we want to add criticals, a 1 breaks the lock; mostly okay, with the long acknowledged fringe problem of experts being incompetent 5% of the time. What does a natural 20 get? I adore opportunities to be creative, but there’s not a lot better than, “You did it perfectly.” A regular success earns that according to the rules, I don’t want to take it away. A speech about how cool and ninja the PC is can come off pretty cringey to me. The correct mechanical answer would be to let the 20 roll over to the next check because the PC’s in the zone or whatever. Not awful, but it doesn’t directly reward the player right when they rolled the 20, which is the occurrence we want to feel good. We’re also rewriting several rules at this point.

Meanwhile, PF2e baked degrees of success into everything. On a crit fail they break the lock, on a fail they leave traces of their fruitless efforts, on a success they get one success toward opening the lock while scuffing it up a little, and on a crit success they get two successes and leave the lock looking pristine. The players don’t feel cheated when they get a normal success and scuff up the lock. The 20 has some reward for most characters. The 1 has a setback, even a reasonable setback for an expert with a +25 trying to open the DC 10 lock on Grandma’s rickety shed.

I actually don’t mind pass/fail skill rolls in D&D or other games. Rolling a 20 is inherently satisfying to me. But I adore the DC+10 critical threshold for making a good build feel like it was time well spent, in or out of game. And since the natural 20/1 and critical rules are connected at the hip, I’ll gladly take them both.

...If we fall off the rope bridge because you did a backflip I'm haunting you though.