System Musings: D20 Skill Checks
I've been spending some time reading and thinking about Pathfinder 2 and decided to share some random thoughts on its design and how it relates to other versions of D&D. This will by no means be an exhaustive examination of D&D-like games. The focus here is firmly on Pathfinder and modern versions of D&D (3e onwards). First up are skill checks; other topics may follow.
Although Pathfinder certainly deserves to be judged on its own merits, this is not a product review. Here I am interested in the comparison between closely related games. Not because I want to stir up edition wars, but because I hope that the contrast will help to illuminate design choices and their effect on the games in question.
One aspect that I find particularly interesting when considering Pathfinder's second edition is how it has evolved from D&D 3.5. Pathfinder started as a variant of D&D 3.5, but now Paizo had a chance to iterate on the rules. No doubt the Pathfinder designers had a good look at the D&D editions since 3.5 when approaching their version, and that is the context in which I'm considering it.
There is no question that Pathfinder 2 is more complex than D&D 5e, and that is clearly by design. So it may come as a bit of a surprise that as far as skill checks are concerned, Pathfinder's design is more streamlined in many ways.
Skill DC, a Passive Skill by any other name?
One notable feature of skills in Pathfinder 2e is that each one explicitly comes with a Skill DC, which is simply 10 + all relevant modifiers. This suspiciously looks like passive skills, which first appeared in D&D 4e and are still used in D&D 5e. There are some differences in how they are used, and Skill DCs are a lot more central to the PF 2 skill system than Passive Skills ever were in D&D.
Passive Skills have been discussed a lot for both editions of D&D in which they appear. I will focus mainly on the 5e incarnation of the concept because I'm more familiar with that edition. Here is what the rules say (from the D&D 5e SRD):
A passive check is a special kind of ability check that doesn’t involve any die rolls. Such a check can represent the average result for a task done repeatedly, such as searching for secret doors over and over again, or can be used when the GM wants to secretly determine whether the characters succeed at something without rolling dice, such as noticing a hidden monster.
You'll note that both of these examples will usually compare the PC's Passive Check against a fixed DC, like the DC to notice a trap, secret door, or the monster called out in the text. The problem with this is that the outcome is essentially pre-determined. A prepared GM will know well in advance whether the party will pass a Passive Check or not. This seems to defeat the purpose of having a check in the first place. The latter case of spotting a hidden monster could actually involve the monster rolling a Stealth Check against the PC's Passive Perception. That would work better but clearly isn't what is meant here as it wouldn't achieve the stated objective to "secretly determine whether the characters succeed at something without rolling dice" (emphasis mine). One other related use case is mentioned in the rules when discussing a player character attempting to hide. Here, the player rolls a Stealth Check for their character, and the result is compared to the Passive Perception of anyone who might spot them. That use of Passive Checks I like. It saves a lot of rolling for the GM and retains a random element. This was also a common use of Passive Checks in D&D 4e. Incidentally, it is also how Pathfinder's Skill DC works.
The main difference is that PF 2 explicitly introduces a Skill DC for each skill and isn't shy to use them. The rules commonly call for rolls against a Skill DC, making it more integral to the system. And there is a good reason for that. The use of Skill DCs allows PF 2 to remove opposed rolls entirely.
No opposed rolls
Pathfinder 2 has done away with opposed rolls. Whenever a skill check is required, it is against a fixed DC. The outcome never depends on a second roll. Traditionally, opposed rolls are called for when two characters directly measure their skill against each other: Sleight of Hand against Perception, Insight against Deception, and so on. This makes a lot of sense. Both characters participate in the contest, and both skills should factor into determining the outcome. But there are some problems with this. The first problem is that this can require a lot of dice rolling. If the GM and one of the players make one roll each, that isn't so bad (albeit slower than just a single roll). But what if more than two characters are involved? Opposed Stealth checks are notorious for this, and it can quickly get to the point where the game gets bogged down in dice rolling. D&D has addressed this particular situation with Passive Checks, as we discussed above. But there is a second, slightly more subtle issue. Opposed rolls tend to be less predictable because the die rolls introduce more variance and, maybe, more importantly, the difficulty is harder to judge for the GM.
Consider a simple example. A player character with a total bonus of +6 to their Deception check is trying to talk their way out of trouble. The guard they are talking to has a total bonus of +4 to their skill to see through the lies (this would be Perception in Pathfinder 2 and Insight in D&D 5e). In PF 2, would resolve the situation with a Deception check against the NPCs Perception DC of 14. It should be apparent that the player needs at least an 8 on the d20 to succeed here. Most GMs (and indeed other players) with some experience in d20 games will have a good sense of the chances of success and, if they care about the exact numbers, will not find it too difficult to figure out that it is 65%. But what about opposed rolls? The PC's bonus is two points higher than the NPCs, so this really comes down to adding +2 to the player's roll and then subtracting the GM's roll. What are the chances for the player to succeed now? This is a lot harder to calculate in your head. It turns out that the player character now only has a 57.25% chance of succeeding (explore this further on AnyDice, if you like). The problem isn't so much that the success chance is lower in the opposed version; it is that the actual difficulty is a lot less transparent.
What I like about Pathfinder's approach is that it not only avoids the issues with opposed rolls but, more importantly, ensures that there is always a roll involved when the outcome is uncertain. No comparing two static numbers. It also means that the system is more consistent overall because combat already works like this with Armor Class as the DC to hit an opponent.
The way skill proficiencies are handled varies quite a bit between the various editions of D&D. The third edition of D&D introduced a substantial list of skills for all characters classes with the ability to raise individual skills through skill points. This system made its way into Pathfinder 1e, essentially unchanged. Allocating skill points gives players some choice in shaping their character's skills, although character class determines a subset of favoured skills. The maximum skill rank is limited based on the character's level allowing for a steady progression and encouraging points to be spread across a few skills (not too many, though, lest you fall behind the curve). This system creates a situation where the modifier for a characters best skills roughly corresponds to the character's level (plus attribute and situational modifiers). Still, a player can spread their skill points too widely and eventually struggle to succeed at what should be level-appropriate challenges.
Starting from D&D 4e, this changed by restricting player choice in favour of a more straightforward skill system. The level-based progression in skill proficiency is now baked in as everyone gets a bonus equal to half their level. In addition, a character is either trained in a skill or not and receives a flat bonus if they are. This system is a lot less nuanced, with little opportunity for character customisation via skills. In 5e, the approach to skill proficiency has been tweaked slightly to reduce the size of the skill bonus but remains very similar. Untrained skills no longer provide any bonus. The proficiency bonus has been combined with the increase by level (but now only increasing once every four levels rather than every other level).
Pathfinder 2, building on the 3e system, seems to have taken inspiration from both 4e and 5e. Here, too, the skill ranks of earlier editions are gone. As in D&D 4e, skills progress automatically with character level, but only if the character is trained in the skill (like in 5e). Skill proficiencies are a bit more nuanced, with four levels of proficiency that provide increasing bonuses. Players still get to choose which skills to increase, but only once every other level.
It is clear that D&D mostly removed the need for players to worry about skills. You figure out which skills you want your character to be proficient in and engage autopilot. Yes, there are some wrinkles to that with exceptions that allow you to gain double or half the proficiency bonus for some skills, but those are special cases, not the rule. I like games with meaningful skills, and D&D 5e doesn't feel like it does that very well. Sure, your character will be better at some skills than others and, indeed, may be better at it than any other party members, but that doesn't mean that when it comes down to a roll, they will actually perform better. Thanks to 5e's Bounded Accuracy and "anyone can attempt anything" design philosophy, the Barbarian can outsmart the Wizard.
The PF 2 system may provide a reasonable middle ground with some player choice on skills every few levels. In practice, however, the decisions are mostly made during character creation, where you'll have to decide on a few core skills to focus on and then raise them whenever you can. That isn't very interesting once the initial choice is made, and precisely why recent editions of D&D have automated the process. The initial choice still matters, of course. Being trained in a skill unlocks new actions, and increasing proficiency enables the reliable use of different strategies in and out of combat.
There is more I have to say about skills, but this post is already on the long side. I'll save the discussion of modifiers and degrees of success for another time.
So what is the take-away from all this? There are some notable differences in how skills are handled across D&D editions, and these are games directed at different audiences. Tastes vary, and you may enjoy playing one more than the other, which is great but not the point of this discussion. I'm not interested in edition wars or deciding which game is best. What I'm interested in are the nuggets of design wisdom that can be gleaned from the comparison. So what did I learn from this exercise?
I thought that the way PF 2 embraces passive skills to turn them into a general mechanism to determine the DC for what might otherwise be opposed checks is very interesting. It makes for a streamlined, more consistent design with fewer exceptions. As I explained above, I think that opposed rolls have their problems, and Pathfinder certainly avoided those in an elegant way. My only problem with this design is that it often makes players passive participants in a contest. This isn't a new issue, of course. It happens every time a PC gets attacked. There are good reasons for this approach, but I can't help but feel that the sense of agency is diminished somewhat. What would an alternative look like? I would be inclined to try a design where the players make all the rolls, ensuring that they always have a hand in their character's fate.
When it comes to skill proficiency, the different flavours of D&D offer different tradeoffs. I don't feel like any of them is entirely satisfactory. I want to like Pathfinder 2's approach, but although it provides a compromise in terms of granularity, it doesn't succeed at offering consistently meaningful choices as you level up. I wonder whether it would work better if skill increases were tied to feats. That would certainly require some changes to how character progression works, and it probably wouldn't change the fact that you are usually better off focusing on a small number of skills. But if instead of saying "Yep, I'll increase my favourite skill again", it was a choice between at least a handful of feats, each of which offered a different benefit alongside the skill increase, there might be some interesting tradeoffs.
 Purists may point out that, technically, D&D 5e doesn't have skill checks, only Ability checks. That is true, at least as far as the rules-language is concerned. So how could anything possibly be simpler than that? D&D 5e's design doesn't eliminate skills entirely, and rolls involving skills still have to be resolved in play. It is this interaction with skills at the table that, I think, PF 2e has streamlined better in some ways. ⮌
 Although they are related to the concept of Taking 10 from 3e Passive Skill Checks aren't quite the same. ⮌
 There is a slight additional complication here because, in the static DC example, it is enough to reach the DC, i.e. a 14 is a success. In contrast, the opposed roll requires the player to exceed the opponent's result. ⮌
 Although PF2 does not slow this down one bit compared to 3e or PF1. There is an increase at every level. ⮌
 To be clear, I'm not saying this is necessarily a bad thing on the whole, just that it has consequences for skills and their use that I don't necessarily like. ⮌
 Doing that based on a discussion of how they approach skills alone would be dubious at best anyway. And even if that made sense, there are so many great games out there that comparing a few flavours of D&D and saying, "This one is the best!" would be pretty meaningless. ⮌