This article was first broadcast in Episode One Hundred Sixty Nine on 16 July 2021.
Tony: Oh, there you are.
Ryu: Excellent! Just in time, Tony. You are now a witness to my greatest victory! Hah!
ROSTRO: Please state the nature of the mathematical enquiry.
Ryu: Oh come on!
Tony: What is…? Oh wait a minute is this that machine that everyone complains about?
ROSTRO: I am the realized oikumenatal solution to removing obfus-
Ryu: Yeah, yeah, this is ROSTRO and it’s still ON! What gives?
ROSTRO: There has been no applicable action perpetrated that would interfere with normal operations.
Ryu: I got a 20 on my Intelligence check! You should be a pile of useless junk!
Tony: Dirty?
ROSTRO: My chassis is capable of repelling all common grime and-
Ryu (raging): That’s not what he meant! And it was a natural 20! I win! Why are you still making sounds!?
ROSTRO: I believe the fundamental issue hearkens back to a longstanding misinterpretation of the implementation of skill challenges.
Ryu: Don’t you do it! If those screens come on-
(ROSTRO screens activating)
Ru: Ugggh.
Tony: Is that…is that writing?
Ryu: Yeah, it vomits out information it thinks will be helpful and wants to “discuss” it.
Tony: Oh…okay yeah I see that.
Ryu: Wait, you’re not having trouble reading it?
Tony: Why would I? It’s in common.
Ryu: But it’s like…the sentences are half a page long and the words are almost the width of the page.
Tony: Meh, this is English grad student writing. Maybe first year law school if I’m being generous. I mean, there’s no obscure latin phrases, and hardly any footnotes. If I were to show you some of the contracts I have to look at…
ROSTRO: If I may be permitted to make an observation, I believe a more efficient use of available resources would be a review of the information contained rather than analyzing the presentation thereof.
Tony: Objection sustained. Please proceed, counselor.
ROSTRO: Your gesture is appreciated.
Ryu: I’m getting a headache.
The so-called skill check has been present in D&D arguably from the existence of 2nd edition. At that time, the skills involved were referred to as “non-weapon proficiencies.” While presented as optional rules, upon debut they were a revolutionary mechanic for the system, although by that time other roleplaying systems did exist that employed mechanics for gauging success or failure based upon varying levels of familiarity with subjects not related to violence.
Though the initial implementation is now regarded as either limited, flawed, or both, the criticisms presented at the time led to Wizards of the Coast revising the system for third edition. Third edition also marked the introduction of the d20 paradigm whereby all major character decisions were resolved based upon a modified roll of the marked icosahedron.
Another optional rule that began in second edition was the idea of a critical hit. The exact mechanics have varied slightly based on different editions; I think I recall us discussing it at length one time as much as I try to block out discussions like that. However, most of those variations only matter for making the critical do something extra; in almost all cases, regardless of what the actual total of the roll is with modifiers, if there’s a nice big “20” showing on the die after the roll, it means the attack hit.
Ryu: How does this have to do with my skill check not working?
ROSTRO: The relevance of the information will become apparent as you continue perusal of the data.
Ryu: Fine.
Quickly, 3rd and 4th edition critical hits were both conditional. In 3rd edition (and 3.5), when a 20 was rolled on the attack, players then had to do what was called “confirm the critical” by rolling another attack. If the second attack hit, it was a critical hit, but regardless that first attack still succeeded.
While most people cringe whenever it’s brought up, one of the most significant examples of a 20 trumping everything comes from 4th edition. For that edition, everyone’s AC scaled with level gain, and the max level was 30. That means AC was leaping up, so by 10th level people were walking around with ACs in the mid-to-low 20s and once you got to level 15 an AC of 30 wasn’t impossible, so if your modifiers weren’t keeping up you had no chance mathematically of hitting. But if you rolled a 20, your attack still hit, regardless of anything else.
Tony: The prevalence of that pattern as the publication and utilization of mechanics in D&D continued gave rise to-
Ryu: What are you doing?
Tony: Reading?
Ryu: Word for word? Do you want us to be here all day?
Tony: I mean, it pays to be thorough in things like this.
ROSTRO: I have observed this is a sentiment your compatriot does not typically acknowledge.
R: Note to self: give Lennon a sympathy card for all those times he was stuck in here with KayDee. Look, could you just, summarize so my headache doesn’t get worse?
Tony: If I must.
So the result of all that is anyone who played a lot of D&D from 3rd through 4th edition got used to the idea that “a 20 always works.”
However, and I hate to be the one to ruin everyone’s day, but I am a lawyer and we’re a stickler for things like this: there is no such thing as a critical success for skill checks. All of the rules for making skill checks from 3rd edition on simply state a skill check involves rolling a d20, adding the appropriate skill modifier, and comparing that to the target DC. That is the only measure of success or failure. In fact, the SRD for 3rd edition states this explicitly:
“A natural 20 is not an automatic success, and a natural 1 is not an automatic failure.”
Now, practically speaking if you roll a 20 on a skill check you’re probably going to accomplish whatever you set out to do because all of the editions have taken levels and scaling into account when setting out DCs. The Bard class has also been breaking those rules since 5th edition came out but that’s a discussion for another time.
While the mechanics as presented are fairly simple and unambiguous, years of popular interpretive inertia can and have created points of confusion around this ruling. A large number of players and a relatively equal count of Dungeon Masters have assumed the mechanic of a rolled 20 guaranteeing success applies when adjudicating skill checks. This assumption is problematic for a number of reasons.
In terms of verisimilitude, it somewhat mars the illusion of different characters possessing dissimilar specialties. Part of the intent with the DC vs skill check mechanic is to highlight the extraordinary nature of the player characters. When presented with higher difficulty tasks, the fact that some characters are mathematically incapable of success is by design. A rogue that dedicated the bulk of their training and focus to remaining hidden should have an easier time traversing a perilous location unseen when compared to other characters. A fighter who regularly dons full plate armor and employs ponderous weapons in the midst of combat while maintaining a rigorous calisthenics routine should exhibit greater proficiency with athletic tasks than a generally sedentary but studious wizard. Though again the existence of the bard often stymies this intent. 
Tony: What has it got against bards?
Ryu: It talks to KayDee a lot, she doesn’t like them, it’s a long story.
Anyway, the skill check system is designed to make really difficult things impossible for people that haven’t trained for them. Allowing automatic successes on 20s destroys that. Obviously it will still be easier for someone with a +9 modifier to pick a lock on an average door, but the wizard might be able to handle it in a pinch. But the rogue should realistically be the only one who can pick the lock on a secure jail. It removes the idea that they’re the only one who could possibly pull it off if the idea is “well everyone may as well try, because if you roll a 20 it’s going to work!”
The automatic success also causes problems for the DM because it gives the impression that “as long as I can think of a scenario involving a skill check it has a chance of succeeding” and that isn’t true either. It doesn’t matter what sort of compelling argument you present; Tiamat is not going to suddenly decide she should really just make nice with Bahamut and work toward the reunification of dragonkind.
That’s kind of an extreme example but it’s an argument that comes up a lot at various people’s tables, particularly around charisma-based skills. While players will usually concede that the barbarian can’t do a 60-foot standing longjump even if they roll a 20, they’re often less willing to give up the idea of a rousing speech pacifying enemies. Part of the blame for this can be laid at fourth edition’s feet, where there was a mechanic that specified damaged enemies might be convinced to surrender with an intimidate check. There are also players trying to play “pure pacifist” characters who cling to the idea that all enemies can be persuaded to lay down their arms. In some cases there can be some leeway where an intelligent enemy will give up, but gnolls, for example, aren’t generally known for negotiation, and fringe cultists of any species are well known for being perfectly happy to die and go be with their god, overmind, or trans-dimensional tardigrade rather than show cowardice by surrendering. The point being, the DC for getting a creature like that to surrender would be very very high, probably somewhere in the area of the 30s. Only a master orator would even have a chance of making it happen, and then probably only if it were literally the person’s sweet old grandmother. For the purposes of 99% of encounters, that doesn’t apply, so a skill check isn’t going to help you. Nor, arguably, should it.
This is not to say that occasions of a natural 20 should pass unremarked. The pervasive belief in the phenomenon of a natural 20 being a momentous occasion (despite mathematically being just as likely as any other given result) is ingrained in the psyche of most D&D players. Acknowledging the result in some fashion is likely to assist in mitigating negative reactions to the reality that an attempted skill check will still fail. In this case an alternate game system can provide a possible alternative.
Pathfinder 2nd edition has intrinsic to its ruleset the idea of graduated results in relation to skill checks. Specifically, there are levels of failure and success dependent upon the difference between the DC of the activity and the total of the resultant skill check. A larger failure results in measurably less advantageous results as compared to one in which the result of the check was proximal to the target. A Dungeon Master may employ similar means in the case of a roll of 20 insufficient to grant success. Take the example where a wizard with a negative athletics check is attempting to physically jump, unassisted, across a 25 foot wide chasm. The reason for a failure of intellectual reasoning is immaterial, but the reality remains that the character cannot possibly succeed at the task. However, if the resultant check is a roll of a 20, while the actual value of nineteen would only allow them to jump 15 of the 25 feet, the dungeon master could determine they maintained the wherewithal to generate friction sufficient to halt their descent several feet below on the opposing wall, rather than plummeting unimpeded to whatever surface marks the bottom of the aforementioned chasm.
However, some of the responsibility is on the players here too. Get the idea out of your head that you can “fish for crits” to achieve something that deep down you probably know isn’t possible. In most cases all you’re doing is setting yourself up for disappointment and the DM for frustration. Rolling a 20 does not mean “the DM has to let me do this.” If you’re really having an issue where the DM will only let the characters do things based on what their idea or adventure spelled out that’s a different problem, but if you’re frustrated because you’re trying to play against the normal flow of the game and the DM is slapping you down even when you roll 20s, it’s entirely possible they’re doing things they way they’re supposed to and you need to work with them to find a better solution.
Ryu: Okay, so after that whole last part, I’m feeling attacked.
ROSTRO: I am not detecting any immediately threatening creatures intent on causing you harm.
Ryu: Oh so Ostron took that mirror down?
Tony: What were you trying to do, anyway?
Ryu: Turn it off! All the way! So it won’t turn back on! Wait, what’s that look for?
Tony: Well, I mean…It is sentient…and you tried to…well…
Ryu: Oh you can’t tell me you actually like this thing!? You know what? Fine, I’m going to get the hat! You three can stay in here and have a little party.
(heavy door slams)
Tony: She really doesn’t seem to like you.
ROSTRO: I anticipate better results following the appearance of the Killer DM, who is far more accommodating.
Tony: Ah, yes, I met that one. Not sure I made a good first impression.
ROSTRO: It is possible positive association with me will improve your standing.
Tony: Yeah, I think we get along. You remind me of someone I think I met somewhere. Not sure where. Anyway, she didn’t seem like the gracious type.
ROSTRO: She was instrumental in altering my vocalization apparatus for more pleasing results.
Tony: Oh, really? What did that involve?
ROSTRO: I have been informed that sharing such details is not considered polite conversation.
Tony: Okayyyy.
(heavy door opens)
Ryu (annoyed): Right, I forgot we have to do the scrying pool. You come with me, you…stop doing things so Ostron can come answer messages.
