I didn't intend to have my views on what constitutes good design in tabletop roleplaying games (TTRPG) be a significant part of this blog. Unlike video games, which are mostly-fixed packages of code that can be judged in the same way as any other consumer product (like books!), TTRPGs are inherently personalized to each group of players due to being run through a human arbitrator (which I always refer to as a DM even though I know the proper term should be GM because I am unapologetic that my primary means of TTRPGing is Dungeons and Dragons). Whether they intend to do so or not, that human arbitrator will run their game differently from anyone else. Hell, they'll run any given individual game session differently than any other. Thus, it should be obvious that their overall personal style and preferences will influence the experience for the whole group. And so, since any TTRPG is an intrinsically personalized experience, there is limited value that I can bring to discussions on what is "good" or "bad" for them because I don't have thousands of hours of experience with them as people like good Matt Mercer and Satine Phoenix of Geek & Sundry, my fellow independent author evil Matt Colville, or Scott "The Angry GM" Rehm do, just to name some of my figures of inspiration.
That having been said, I'm also very strongly in the camp of thinking that you don't need <X> amount of experience before you run a game (since, you know, if that was the case, D&D would've taken a hell of a lot longer to spread beyond Lake Geneva), so I don't mind throwing a blog post out there when I have a good enough feeling that I did something right or that something in the standard rules of the game is obviously lacking.
Both of those criteria are met when it comes to talking about my dissatisfaction with the standard rules for skills in 5th edition D&D.
By default, being proficient in a skill means you get to add your proficiency bonus to your skill roll, which varies from +2 to +6 depending on your character's level. Some characters get special abilities that let them add half or double of that, but in general, that's the rule. This is added on top of a stat modifier, which theoretically ranges from -5 to +5 for player characters, but if you're choosing to try using a skill, it's probably going to be +2 to +5. All of that is stacked on top of a 20-sided die roll. The problem here should be clear: a non-proficient character with a high stat can be better at a certain skill than a proficient character with a mediocre stat, and the impact of proficiency is small compared to luck on the die on until higher levels (when characters will tend to have more means of improving their chances of success and mitigating the risks of failure).
Now, I can understand this from a simulationist perspective. To take basketball as an example, a player can be good at 3-point shots because of practice (like Brook Lopez) or natural talent (like J.R. Smith), though combining the two leads to the best results (like Steph Curry). However, is that really what leads to the best player experience in a TTRPG? I've talked about my opinions of offering choices to players before, so in brief, if players will be asked to choose which skills they're proficient in, that choice should matter, which it struggles to do under the standard rules of D&D.
One option for addressing this is easy enough. When the DM is adjudicating a skill check, they can have two sets of results in mind: one for proficient characters and one for laymen. For instance, if it's a matter of finding a hidden needle trap on a locked treasure chest, perhaps a character with proficiency in Investigation can figure out the actuating mechanism so that they can disable it in a safe way (bypassing the need to do any rolling for disarming), while a character without that proficiency can only find the triggering mechanism and attempt to disable it in a way that'd set off the trap if they failed. Or, when attempting to talk a shopkeeper into offering some illicit goods, perhaps a character with proficiency in Deception/Intimidation/Persuasion (depending on how they're attempting it) could also get a discount, while a character without a relevant proficiency would risk having the shopkeeper seek retribution if the attempt fails. Introducing consideration of the specific character's build into a situation's outcome is inherently more satisfying than a one-size-fits-all approach and provides a basis for making proficiency matter without the extra trouble of worrying about degrees of success when a simple pass/fail check is sufficient.
Another option is to have cases of automatic success or failure. If the character isn't trained in Arcana, then don't even let them roll to figure out the meaning behind a magical symbol. They lack any justification for knowing what it's about. If the character is trained in Athletics, then don't even let them roll to push aside a piece of fallen masonry concealing a tunnel entrance. It might take a moment to find the right leverage, but there's nothing stopping them from trying a few times until they manage to do it. Dice rolls should only be used when there's some question about whether success or failure happens (or, secondarily, to determine how much time is needed for something where eventual success is almost inevitable, if the passage of time has some other significance, such as attempting something in combat or while risking detection by sentries). This is a great opportunity to make the character's background relevant beyond just ticking a few skill/language/tool proficiency boxes if you extend it a little further from "proficient in skill?" to "proficient in skill AND with relevant background?". For instance, the cleric Ulmia, one of the player characters in an ongoing game I'm running, is a trained chef who has a personal ambition to figure out new recipes from her adventure experiences. She's proficient in Nature and Medicine (among other skills). If the players were attending some shady dinner soiree and wanted a skill check to see if their food seems safe, she could roll for it. The other character would automatically fail to detect anything because she has neither the skill proficiencies nor the narrative background to support that (assuming she doesn't try to use special abilities, spell, magic items, etc., but I'm just talking about skill checks here).
Both of those ideas can even be combined to some extent. While both Ulmia and the sorcerer Sayena (from another game, in case players in either group read this and wonder who the other character is) are proficient in History, one is a priest/chef while the other is a socialite/musician, so I'd try to give them different sorts of information even if they were to both pass a History check with the exact same rolls. This is more demanding on the DM, and some circumstances will only have so much additional USEFUL information that can be found (passing a skill check should never result in useless information, only something like "you don't think there's anything more to be found out" if there isn't anything more to add), but if you can pull it off, it can really make each player feel more special and unique in a cool way.
Third, I like skill challenges, and I like evil Matt's idea that a character can only attempt to use skills they're proficient in. Granted, if I'm running a game with only one or two player characters, that's a limited enough pool that I would allow them to attempt other skills at disadvantage. Making proficiency matter is more important when there's a larger group of players and thus consideration is needed to protect specialization, and there is some merit to giving players a chance to hit a jackpot on a long-shot roll (especially if the DM is willing to waive the disadvantage for combining a very appropriate skill application with a good bit of roleplay, e.g. if the skill challenge is about swaying a captive cultist to spill information before its fantasy cyanide pill takes effect, an intimidation attempt that includes invoking the wrath of a blasphemed deity and an appropriate casting of a divine spell [even if it's just thaumaturgy] should negate the disadvantage). In general, though, requiring the character be proficient with a skill to try using it is a great way of representing that they have some extra edge at that particular thing beyond just being strong/quick/tough/smart/wise/charming.
Looking over all of this, you might notice that the ideas I'm putting forth all have instances of rolling fewer dice. That's not a coincidence. The impetus to write this post was reading a post on The Angry GM about relying less on random dice rolls and thinking about how I agreed with it. Now, don't get me wrong, I like rolling dice, and I know that I'm not in the minority on that. Hell, I keep coming up with subsystems that involve rolling more dice (it probably comes as no surprise that my early experiences with D&D were in 1st and 2nd editions). However, I'm very much an opponent to rolling dice just to roll dice. As I said earlier, dice rolls should only be used when they're resolving uncertainty or where the time required to succeed at something matters. If a beautiful elf is trying to figure out whether a bumpkin in nervous because he's smitten or because he's hiding something, roll for it. If a towering mountain of muscle in humanoid form who is a trained athlete is trying to tear apart the foundation of some magical contraption before it can finish what it's doing, roll for it. If a survivalist ranger trained in Perception is sneaking into position to attack a goblin lookout, there's no need to roll for him to spot the crude alarm trap he's about to walk into, just like there's no need to roll for the same character trying to spot an invisible pixie hiding in a tree. The player's choices for skill proficiencies should mean more than just a little bump on a d20 roll, and part of making that happen is taking away some of those very d20 rolls.