Fixing Your Skill System

A mindless integration of skills really, really sucks. And 5e’s skill system is mindless.

I’m talking specifically about D&D 5e, but my objections apply to many other games as well. Skills are interesting, and important, but they’re also one of the easiest things to balls up when it comes to running a table, and a well-run skill check is something that I think many players take for granted. But I don’t hate every aspect of skill systems – just the lazy, mindless ones.

Skills are good because…

  1. They allow players to meaningfully specialize their character.
  2. They make each character’s niche or spotlight obvious at a glance.
  3. They allow the GM to provide information in a way that feels fair but not overly generous.
  4. They give a mechanical framework to challenge players outside of combat.

But the 5e skill check doesn’t do… any of these things. In fact, they’re antithetical to the 5e skill check. So how did such a simple idea as Roll+Mod end up so far off-target? I’d argue that skills suck because both parts of the equation suck – the Roll sucks, and the Mod sucks. And here’s why.

The d20 roll is a wide-variance result – no matter the character, you’re equally likely to get any option, and the extremes are quite far apart. To contrast this, the mod is tiny. It has a limited, narrow effect on the roll. Some things that can actually happen at real tables populated by smart people trying to have fun:

  1. The ranger (15 WIS, an Outlander, with Survival proficiency) tries to track an animal. The cleric (16 WIS, no proficiency) who has never stepped foot outside their underwater monastery, attempts the same. In 95% of roll-cases, these two characters are equally effective.
  2. The arcanist rolls a Arcana check (getting a 12, vs DC 15) so the GM shrugs and says that they don’t recognize the obviously magical flour. The GM shelves away their cool baking system. 
  3. The players have to sneak past a guard. The thief is the only person with decent stealth, so they go ahead. For the next 15 minutes, the rest of the players twiddle their thumbs. 

So what are the problems?

Well, firstly, proficiency is so negligible as to be laughable. 5e’s movement to smaller numbers is lovely, but the perpetual marriage to the d20 means that these small numbers only offset the whims of a capricious die. For the first 4 levels of the game, proficiency doesn’t matter on 90% of d20 rolls. Even at max level, it doesn’t matter on 70% of rolls. If your bonus is +2, there are only 2 results of the d20 where the bonus changes a fail into a success – 10% of roll-cases.

Secondly, the system is so worried about player’s being able to face each challenge that it is unwilling to restrict options. This is generally true across modern “freeform” or “broad” systems, any system that tries to simulate a wide range of activities; anyone, no matter their positioning, can take the shot. Sure, you’ve never picked a lock before – but you can try!

Finally, the very concept of a “knowledge check” makes my skin crawl. The only intelligible rational behind the knowledge check is that you are determining if a character happens to have learned the thing, and so if you roll above the DC, then the thing happens to be a random part of their education; the difficulty of the check represents the random nature of education. But each knowledge check only has 2 results:

  1. The player learns something.
  2. The player doesn’t learn something.

And what’s more fun – doing things, or not doing things? Progression, or stagnation? Competence, or incompetence? If a player has a high modifier and rolls poorly, then they apparently don’t know common info that their character reasonably or logically would (being incompetent in your area of focus is generally not fun). If a player has a low modifier and rolls well, then they apparently know something that they reasonably or logically would not (succeeding at something via pure luck can be fun, but only in moderation).

And how do we fix them?

Maybe you’re already fixing the skill system. You’ve installed degrees of success – beat the DC by 5 and find more info – or a binary can-they-check condition – you must be proficient to track this animal – or even a dash of narrative positioning – Mandy that doesn’t make any sense, you’ve never even see tracks before. Don’t roll it. Or a dozen other tricks to fix these glaring omissions. Playing Read As Written (RAW), however… them’s the breaks. And judging by the new-GM advice threads I see online, plenty of people are still playing purely RAW.

So how would we modify the game to make the skill system actually achieve those original goals? Those hints above are real suggestions, but here’s a more thorough breakdown of what I do to keep skill checks meaningful and interesting:

  1. They allow players to meaningfully specialize their character. (Proficiency is required for many checks. Additionally, it allows a player to automatically succeed on many baseline difficulty checks)
  2. They make each character’s niche or spotlight obvious at a glance. (Proficiency counts as an additional success on group checks)
  3. They allow the GM to provide information in a way that feels fair but not overly generous. (Provide the information that is reasonable+fun; fairness is not inherently desirable)
  4. They give a mechanical framework to challenge players outside of combat. (Challenge players in the narrative – succeeding only because of luck is not consistently fun.)

To elaborate further –

  1. In order to make specialized or high-difficulty checks, I require proficiency. If you want to know the foreign affairs of the Elvish Conglomerate 400 years ago, take History. The downside to this is that a character’s specific knowledge might not qualify for a whole skill proficiency, but maybe they know a ton about this particular topic. To accommodate that, I allow narrative positioning to trump this rule.
  2. If you have proficiency, you can succeed on checks with DC roughly equal to 10+your bonus. I play this one fast and loose – proficiency is empowering to players and lets them succeed on normal-ish skill checks. If someone invests in specializing, I don’t want the random nature of the d20 to screw them on a simple task.
  3. When making group checks (especially stealth, but it comes up on occasion for Survival, Ath/Acro, and social skills) Proficiency counts as an additional success. This counts as an extra player for the “half of players must succeed to pass the check” existing group check rules. In practice, it means that a +14 Stealth rogue can often get their party through some skullduggery with much less luck than normal. Splitting the party is often loathsome.
  4. Succeeding because a lucky number told you so is often exciting and memorable, but only in moderation. I try to challenge characters outside of combat in ways that don’t rely on the dice – narrative challenges, tests of their reasoning or problem-solving skills, and so on. 90% of the stuff on their sheet is designed only for combat, so I turn to the stuff in their heads for most everything else.


And that’s all. That’s all I do to fix the problems I have with skills. Degrees of success make specialization more relevant and visceral. Checking proficiency as a pass/fail system makes it matter more than the paltry bonus. Auto-successes let players feel good about their character’s specialty without trivializing harder obstacles. Proficiency counting as an extra successful player in group checks incentivizes the party to stick together. And challenging characters outside of combat in a more immersive way than “Roll on the slot machine of learning and see how much fun you get to have today” helps to hide the weakness of skill checks as a catch-all solution.

Leave a Reply

Your email address will not be published. Required fields are marked *