The Role of Rolls

The essence of D&D, and arguably of tabletop roleplaying at all, is randomness. For years it was right there in the name: D&D ran on the d20 system, where players roll a 20-sided die and use rules to interpret the result. Everything about the system comes down to rolling somewhere, and even non-random choices like “you can Hide after making an attack” are actually ways to affect the randomness (here, creating a random chance to hide where normally no such chance would exist). It feels a bit silly to critique 5E for a mechanic it shares with (almost literally) every other game.

The issue at hand is the amount of randomness. Obviously there’s a sweet spot. Too much randomness and there’s no player agency, but too little and there’s no point to the system at all. Which, as I write it, is a succinct way of stating my point: the more randomness there is, the less say players have over what happens, and vice versa. Now that I’m nearly twenty sessions into 5E I think I can safely say the randomness is far more than in any system I’ve played, and that’s incredibly bad.

I participated in the 5E playtest* when it ran more than five years ago. In the rules as they stood when I ran my playtest session, monsters could roll to determine how difficult certain abilities were to avoid. This supplemented rather than replaced the players’ saving throws. That is, a wizard’s fireball DC might be 1d20 + Intelligence, and then a fighter would roll 1d20 + Dexterity to avoid damage. I can understand the tension this creates but in effect it just caused bad play experiences. Neither the DM nor players knew how scary their spells would be until after they were cast, so anything could suddenly be wimpy or devastating. Play dragged as the number of rolls increased. It prevented players from planning for monster DCs, but rather than improving the game by making everything dramatic, it just punished players by also preventing them from learning about monsters over time.

The kicker came when a monster with a fear effect popped up and rolled a 17 on the check to set its DC. All the characters but one fled, leaving a single character facing the rest of the monsters. Another character shrugged off the fear effect one round later, but needed another round to get back to combat (the insidious thing about fear effects—remind me to talk about that one day), and in those two rounds the remaining character died. The other players saw no use in returning one at a time to get slaughtered, and they agreed their day in the dungeon was over and we ended the session.

We ran the monster and used the rules exactly as written, so I doubt this was user error. The game put too much weight on that one die roll by letting it almost single-handedly decide how much of the party didn’t get to play for at least two rounds. Even if I had known how crazy that ability could be, I can’t plan for a 17. Even if I had a contingency for rolling high on that fear effect, what was my contingency for rolling well on a second ability later? What if I had rolled low? What if things went the other way, and the players wasted their most powerful abilities or annihilated everything in a single spell? I understand the fun in dealing with the highs and low of the dice, but at a certain point those dice override anything the players can do about them, and there’s no point to a “see who rolls well today” simulator.

5E doesn’t have this dual-roll system but it does something similar because of how it compresses static modifiers. Attack and defense values, by and large, are not tied to level or CR in any significant way. A 1st-level fighter can have the same AC as a CR 19 pit fiend. This is intentional. The designers wanted CR to be less relevant than in previous editions, and they succeeded. I like that high-level characters can’t blithely ignore mundane goblins and that low-level characters can matter in a fight with a dragon.

But as a result, players have fewer ways to affect that randomness than they did before. By shrinking a static modifier, you lower the modifier importance as compared to the random element. Consider rolling 1d20 + 3 versus rolling 1d20 + 300, and it’s clear the die creates much more variance relative to a smaller modifier. In 5E, for the most part you can’t change your modifiers. At all levels and at all times that d20 is the most important factor in what your character does, more than your race or class or equipment or feats or, most relevantly, any agency you wanted to exert over your character and what they do. A player has far fewer options for luck mitigation, and thus fewer options for changing the flow of the game, than they had before, making everything more subject to the will of the dice.

“But Nameless DM,” I hear you say, “weren’t you just complaining about how players can be so good mechanically that you can’t challenge them? Wouldn’t you want that additional randomness so even the mightiest numbers still have a chance at failure?” It’s a good point, and it’s why I thought I would like 5E more than I do. The issue there is that the rules don’t just take away player agency, they take away DM agency too.

Before I started on this campaign I spoke with three people who had already run 5E sessions to get their advice on encounter building because I wasn’t sure how to properly handle it. This is what they said, paraphrased:

• DM 1: Be careful any time you give players multiple monsters to fight. I had an entire group wiped out by two CR 1/4 goblins before the session could even get going.
• DM 2: Never, never give players a monster with the same CR as their level. It’s going to kill them. Always split the encounter into several low-level monsters instead.
• DM 3: I think the CRs are too low. You know that sidebar in the DMG that says not to send a rakshasa at a group of 8th-level characters? I did that. Not something similar, I did exactly that. Not only did the players win, they killed it without taking any damage.

There’s no common thread here. Each DM had a different experience, and how they dealt with that experience changed their DMing style from then on. If you want a chaotic, anything-can-happen play experience, great. But if you want to have some control over your adventure design, your pacing, resource management, or the emotional impact your encounters will have—really, if you want to be a DM rather than a simple moderator—you’re out of luck.

This is where 5E’s heavy reliance on randomness exacerbates rather than alleviates the Superman problem because it basically eliminates the Dragonball solution. If a character’s AC is too high, there are no numbers to throw at it. The best I can do is give the party a very high-level monster, which can swing its attack bonus by as much as three points. That’s it, just three. But along with that comes a significant increase in hit points and damage, the two things level does seem to affect. If I want to give one character a challenge, my only numerical solution has the side effect of knocking him (and everybody else) near to death with a single attack. We have a phrase for gameplay where characters throw themselves at each other trying to get the one big hit they need to end the battle. It’s called “rocket tag”, and it’s terrible, but it’s even worse when only one side of the battle has rockets.

During the campaign I escalated out of necessity more than desire. I started off with easy encounters to get everybody (including me) used to the system, and the players only had trouble with them when they were especially unlucky. By the last session I was throwing fights at them that by all rights should have been unwinnable and I barely fazed half the party. Nothing I did mattered as much as the whim of the dice, which means I couldn’t actually prepare for sessions at all. I couldn’t even know whether a given fight would last for ten minutes or two hours. That’s pretty critical information when you have a constrained session length and a constrained number of sessions, and I ended up cutting several story threads because we suddenly didn’t have time for them.

I think this is actually the goal of 5E. It’s intended to set up more of a “play to see what happens” experience than in previous editions. But I like giving my groups almost-deadly monsters so they know they earned their wins, I like getting emotionally invested in who my characters are and what they can do, I like seeing players come up with plans I didn’t expect and getting excited about them. I can’t do any of that if every encounter, every fight, every challenge relies entirely on luck, getting resolved almost without the input of the players involved. I can’t set up expectations for the players if I can’t even have them myself.

Now that I’ve been a 5E DM too, I can add my own experience to the advice above:

• DM 4: You can’t build encounters in 5E, not really. You can look at an XP budget and guess at how abilities will work and try to set up pluses and minuses all day, but it won’t matter as much as a couple of die rolls you won’t know about until they happen. Just put stuff together and give it to your players and hope everybody comes out the other side feeling good about what happened.

And I don’t see how that’s fun.

* — Technically I’m not allowed to say this. By the legal terms of the playtest, nobody is allowed to say anything about any part of the playtest to anybody else, in any way, ever, for any reason. I’m assuming those terms have expired now that the material it was intended to protect has been publicly released for a few years. But if I’m wrong, and suddenly this post is about an hypothetical unnamed non-proprietary game, there you go.