Agreed to a degree. It's highly unlikely that a theoretical physicist will make a mistake on a simple algebra equation. Could it happen?
BWAAAA! Dude, you just got a honest belly laugh and a tear from me. You can't be serious, right? You do know that theoretical physicists review their own work over and over and have colleges review their work over and over and then submit to peer reveiw and then whole world review and that, occasionally, even simple algebra mistakes still make it through, right?
I'm an engineer, and I proof my stuff for simple arithmetic errors all the time, much less algebra and calculus ones (surprisingly few calculus errors, as that's most application that creates algebra to screw up). Why? Because I
make them. And I'm a professional.
This idea that someone highly trained in a skillset doesn't constantly make mistakes within that skillset is really, really inhuman thinking. Have you never watched a pro-basketball game? Those are the top 300 or so athletes in the world at that sport, chosen in a brutally Darwinian selection process, and highly paid. They make mistakes at their skill all the time. The hit percentage for free-throws is staggeringly low when you consider that these are literally the best of the best at that sport.
Sure. Your woodworker isn't likely to have a lot of failed simple birdhouses. My point is simply that as you get better at what you do, the threshold of where you're likely to make mistakes rises.
You'd be very surprised. My brother is a woodworker, and turns out furniture worth thousands of dollars at sale. He screws up simple stuff all the time -- but he knows when to toss a piece and start over or if he can still salvage it. He knows how to do the complex and hard things that other people don't, and has the experience to pull it off. He can do things that are amazing with wood. But he still makes mistakes, and sometimes at very easy tasks.
Despite the fact that most of us learned to walk so long ago that we don't even remember it, doesn't mean that we don't sometimes trip (and even injure ourselves severely, occasionally to the point of death). But that sort of random and extremely unlikely event isn't the sort of thing we typically model in a D&D game. So why do we persist with the idea that somebody with a +9 modifier to their skill can somehow fail at something that's a DC of 12, without significant mitigating circumstances? If the base is 10 (that is, your passive score is 10 + modifier). That is, unless circumstances warrant disadvantage (which can still be applied to a passive check), why do we still seem to insist that you can fail by making a bad roll at something that you automatically succeed at if you don't roll at all.
We also aren't often walking in high stress environments. The fall and injure rate during high stress events is much higher than average, and we aren't exactly simulating a stroll in the park when we're playing our elf game. Stress is a big reasons why basketball players make mistakes in games, why are we discounting it as a factor?
Also note that "automatic success" also doesn't mean you don't make mistakes along the way. As you note, it could be that you did make a mistake, recognized it, corrected for it, and ultimately succeeded. That is, success, whether automatic due to a given score, or the result of a die roll, doesn't mean that the process of succeeding was perfect, just that the result was success.
Well, in a six second task resolution window, it's hard to see how a mistake made doesn't cost at least another round of effort.
Which goes back to the original post. If the rogue is automatically succeeding at everything, that's because they aren't attempting anything that's really beyond their capability to succeed. If you think that they are capable of succeeding at too much, then either they aren't meeting enough challenging things, or maybe we're lowballing the difficulty of given tasks. That is, the issue isn't that they are automatically succeeding (although that's another debate that comes up a lot), but that they are automatically succeeding at too much.
When setting a DC, I think the levels of difficulty are based on an individual who is untrained, or has very basic training. It's not really defined though.
Despite feeling these two paragraphs actively fight each other, let's look at the guidance in the rulebook -- DC 20 is hard. DC 25 is very hard. DC 30 is nearly impossible. There aren't DCs over that, and there's a strong argument that they shouldn't exist (not going to relitigate that). If we're stating up DCs based on the average guy, how many DC 25+s are you going to have out there? A skilled, modestly attributed commoner can't hit a DC 25 at all until they get a +3 proficiency, and then it's 1 in 20 times. A first level rogue, well attributed, will hit a DC 25 about 30% of the time. So, yeah, that seems very hard. A DC 30 is impossible for a non-expertise person of even the highest attribute until they get a proficiency of +5, or 13th level. For the expert, it becomes feasible, again at highest attribute, at 5th level, at a 10% chance. So the expert will still almost always fail at that task for quite some time. At best, they achieve a success rate of worse than 50% (at max proficiency).
Enter reliable skill. At 11th, that same character suddenly has no chance to fail a DC 20 check, but still has the same chance to fail a DC 25 and DC 30 check. Reliable? Maybe, depends on the goal.
Reliable talent doesn't mimic actual expertise in a skill.
So picking a lock, for example. I personally think that's a very hard thing to do. Even with a medieval lock which is much simpler in design than today's locks. Would a person without any training and any modifiers be able to pick one? I don't think so. So I'd put picking a lock at a DC 20 at the lowest. That allows a 5% chance off success for even an untrained individual. I'd rather see it at a DC 25, because then you either have to have a very high natural skill, or some training to succeed. But the game isn't currently designed like that. It's designed so that a party of characters will have a chance to succeed at nearly anything. Not necessarily everybody in the party, but that somebody within the party will. We, on the other hand, don't care about that. We're more interested about what makes sense in the game world, and if that means something is too difficult for the PCs at that level, then they can come back and try it later. Or find another way, like breaking the lock, kicking in the door, a knock spell, or whatever. We don't alter the DC for the "designed level of the party," instead it's a question of what makes sense here.
Well, you're wrong. A medieval lock is ridiculously simplistic by today's standards. The real barrier is knowing how to pick locks at all. Once you know how, an average medieval lock is almost trivial. Now, a master's piece lock, well, those were really tricky because many didn't function as key in lock style locks at all, but as puzzle locks.
But, to return to your point, you're setting DCs too high based on your assumptions and not the guidance from the ruleset (which sets lock DCs much lower). When I judge the usefulness of a rule, I do it in the context of the rules it was written to work with, not on how you might interpret the actual difficulty of medieval style locks and your preferred house method of setting DCs.
The benefit of this approach is that you design the probabilities around your expectations. So in the case of picking a lock, we generally think that only somebody with a natural high Dexterity combined with advantage will be able to pick a lock without training. A DC 25 meets that expectation. Somebody with some training and a high Dexterity, or more training with a lower Dexterity can also do it. That DC 25 works fine. A character with a +11 modifier? It's not automatic, but they won't fail often.
Think of it a different way. In all sorts of shows, movies, etc., there are points where the characters have to pick a lock. It is very, very rare for them to ever fail. In most cases it doesn't even matter how much time it takes them. The dramatic contribution of the scene is one that establishes the circumstances of the following scenes. That is, they don't belong there, and presumably don't want to be caught. The drama isn't in the lockpicking skill, it's how the use of that skill frames the subsequent action. Sometimes it's used for comedic purposes, where somebody is taking too long to pick the lock, and somebody else is impatient and just kicks it in. Obviously remaining undetected is unimportant to them.
I generally am not writing a movie script with my games, though. Also, I like using fail forward, so a failure doesn't cause the action to stop, it increases the tension or has a cost. A failure resulting in a picked but broken lock is good in a 'we didn't want anyone to know we were ever here' heist. A picked lock that results is a slip and a loud knock on the door is a good 'we're sneaking in and hope to not alert the guards' heist. A failed pick that still allows for the barbarian to kick the door in is good for, well, places where the barbarian kicking the door in is cool.
Failure doesn't mean the story stops, it means the players have to find a new path or deal with consequences of an unintentional side effect. Failing a skill check, in my games, adds to my game, it doesn't cause it to crash to a stop.
But in the game, we often get hung up on playing the rules. That is, there's a rule for picking a lock, so therefore that rule must come into play. If that rule is "eliminated" due to an ability, then we feel that it's breaking the game. But the skills are really ways to differentiate the characters and their, well, skills. That is, they provide different options, each of which might be beneficial. If you consider the skill the important part, not rolling the dice to see if the skill succeeded, then it changes the entire discussion.
Not my problem, really, as I often allow an approach appropriate to a challenge to succeed. It actually the rules that are inhibiting my ability to set tension moments in this case. But reducing most DCs to automatic, I can't decide that an approach has a chance of failure and a penalty so a check is needed because the rules have taken that away.
Ultimately, this really is my problem -- reliable talent removes things from the list of what can be a tension moment. It limits the possible fiction. And that's my problem, because of my playstyle choices.
For example, the goal is to get into the building. The rogue has a skill to pick the lock. The wizard has the knock spell. The fighter can kick the door down. Any one of them can break a window to get in. And any one of them can search the area and potentially find the key that's hidden nearby. Where's the drama? What makes one option more dramatic than the other? What drama is lost if the rogue can automatically succeed at picking the lock?
If there's no cost to making noise or destroying property, there is no drama, I wouldn't even ask for a check. My players would announce their actions, they'd be successful, I'd narrate the success, and we'd move to a bit that has some drama.
If, though, as it would usually be, knocking or bashing in the door, or breaking a window would alert those in the building, then there's some drama. The rogue can pick the lock quietly on a success, and that's a tension filled moment -- a failure means going loud. But, reliable talent tells me I can't do this without artificially jacking up the DC on a lock put in a kickable door. It's not limiting the tension points in the game because the rogue can pretty much always pick the locks. While I can deal with this, it's irritating.
The reality is, without something else, there really isn't any drama. Just different ways to get inside. Presumably breaking a window is the last option (or not really one at all) because they want to remain undetected. The knock spell, and even kicking in the door are also not desired for the same reasons. Otherwise there would be no debate. But it's the fact that they want to get in undetected that really matters, not the specific method of getting there.
If these are low level characters, they might have to resort to another alternative. But if they are high level characters, wouldn't the expectation be that the rogue will just be able to pick the lock? Unless the lock is especially complex, or whatever, they've picked hundreds if not thousands of locks with the same basic design. Why are we concerned about failure at that point? We've moved past that point in the character's career.
Because I want to retain the ability to determine if the circumstances allow for automatic success (something I'm actually quite generous on, given appropriate approaches), not have the rules tell me that this class of thing can't be a challenge anymore. Reliable talent doesn't make the rogue better at his craft, it restricts the challenges the DM can use in the narrative. Another reason I like the advantage option rather than the floor option for reliable talent. It still allows for a challenge to be failed, if remotely, across a broader range of possibilities while actually improving the rogue's craft against
all possibilities.