In a recent blog post, Sly Flourish talks about 1st level as the high-water mark for danger in 5e. This sparked a conversation about danger and difficulty levels in 5e more generally. I also follow the excellent Jennifer Scheurle on Twitter, who wrote this article on difficulty in video games, particularly emphasizing Souls-like games, for Polygon. In the article, Scheurle unravels the concept of difficulty, and I would like to do something similar for D&D and tabletop gaming in general.
Scheurle explains in detail that difficulty in a video game is about repeated effort at a task and learning to succeed, all of which turns on the player’s trust that the task can be learned and surpassed. Whether it’s fighting or platforming games, you’re largely tested on your ability to press the right button(s) in a variously narrow window of time. If you fail in a PvE game, you lose some amount of progress and try again. (PvP games are, of course, only as hard as your opponent is skilled.) You fail, then learn, then succeed – setbacks are part of the experience. The frequency of setbacks is going to be a huge part of the player’s perception that something is “difficult.” Ironman modes show up in a lot of games, but especially those where the setbacks involve losing assets you can afford to lose – there’s no main character whose death ends the whole game. (Yes, there are exceptions, we don’t need to drill down on this right now.)
These ideas both do and don’t apply in tabletop gaming, because we handle task performance and failure so differently. There are no saved games to load. It’s always in Ironman mode. There’s no timing of a button press, just dice to roll. (Diceless games are a question for another day.) A button press is difficult if you have a fraction-of-a-second window to complete it. A roll is difficult if you only succeed on a small number of results. A player doesn’t learn from bad rolls – no matter what dice superstition tells you, you can’t get a better roll through practice.
As a result, we introduce things the player can do to tilt the odds in their favor. In editions earlier than 5th, you might be playing the positioning game to pick up a +2 bonus to your attack roll. In 5e, we’re talking about gaining advantage or rerolls on failed attacks, reactions like shield, and so on. You start with tightly limited resources that gradually expand over the course of play – everything from standard available actions like Help to phenomenally powerful effects like foresight. The skill of the game is about making judicious decisions, rather than precise timing.
There’s a lot more going on in combat than just hitting and missing, of course. Making good tactical decisions is an area of skillful play that video games and tabletop games essentially share.
When you lose a fight in a tabletop game and it isn’t a TPK, your party survives somehow and might be able to hurl themselves at the same fight later. Even if they do, though, the narrative context is different: any creatures they killed the first time around may still be dead (unless the enemies have raise dead as well, or it’s just a bunch of zombies to replace). Recovering from that loss probably cost the players something, even if it’s just a couple of 500-gp diamonds and the time-lapse of a long rest. The enemies also respond to you differently, because it’s their second time facing you as well.
What’s Up, Danger
That brings me to Sly Flourish’s article, linked above, and the conversation on danger and deadliness that sprang out of it. The gist of the conversation is that the combat-by-combat risk of death is low by 2nd level, and generally goes down from there, thanks to increasing hit point totals and spell slots. The logic works out, even if my own campaign has been a bit of an exception. I’ve had one 1st-level death (a bleedout accelerated by a nat 1 on a death save) and two at 5th or 6th level. One got critted on by an Oathbreaker paladin and then swarmed by ghouls, while the other drew too much attention in a fight that was supposed to be too hard to take on (but the PCs tried it anyway); the PCs won and got him revivified.
My campaign does the things that Sly Flourish suggests to mitigate danger at 1st level: +5 starting hit points, and you almost certainly level to 2nd after the first combat encounter. I feel two ways about 1st level being the most dangerous, even if I don’t lean into that myself. The bad side of that is obvious enough: characters are dead before they’re even established, you might be chasing off new players with a bad experience, and you don’t even feel that great about the death because you died to the schlub monsters. You can’t even say you’ve experienced your class and subclass until 3rd level or so.
There would be no shame at all in starting campaigns at 3rd level.
On the other hand, as long as it’s not a TPK, a defeat or character death can be great for a campaign. It establishes the stakes and shows that the DM follows through with what gets established. This is your regular reminder that one of the worst things that can happen to a narrative is backing off of the stakes that got established, after the fact. Don’t defang losses, don’t hamstring victories. If you want players to know that losses are part of the campaign and there will be setbacks before victory, you must establish it early. Ideally, within the first three sessions.
I also want to point out Court of Swords as an example of an insanely dangerous campaign even at high (13th-14th) levels. Season Seven has multiple character deaths and even more close calls. (High-level half-orc Zealot barbarians are just unbelievably hard to kill.) Part of that is a willingness to divide and conquer, part of it is fantastically unkind dice, and part of it is simply big, deadly monsters that aren’t nearly as designed-to-lose. The DM is very good at making sure the story can go forward in some sense even when the characters fail.
High Difficulty/High Danger
The thing about high difficulty is that a lot of DMs are bad about creating situations where a failed roll stops the action and the interest. For at least as long as I’ve been gaming, there has been game-running advice about making sure that the action doesn’t hinge on a single roll succeeding. The better implementation of high difficulty is to frame failure as the costly way to succeed. Failing a significant skill check is a step toward a catastrophic failure, but you can go forward for now.
The other way I see to increase difficulty in combat sequences is to add more environmental effects that deal damage or impose conditions if you’re standing in the wrong place. This is an obvious lift from MMO boss design; it’s not a coincidence that Stands-in-Fire named his blog that. He has an excellent understanding of how to add moving parts so that players have to think and move throughout the fight.
These make the fight more difficult by adding problems you (probably) can’t stab and adding more decision-making factors. This kind of mental load is more characteristic of 4e than 5e. 4e was well-known for presenting combat as a puzzle to be solved, and it’s reasonably evident that more moving parts make puzzles harder.
High danger is another matter. Look, DMs play with the ultimate stacked deck. In all encounters, the players can win only because the DM has, at some level, decided to allow their NPCs to lose. (We’ve all heard or experienced horror stories of DMs who had decided that this NPC could not possibly lose.) If you want more danger, you can always have more bad guys show up, or have more bad things happen. The conversation about making 5e more deadly requires us to take the XP Thresholds By Character Level table of the DMG as proscriptive rather than descriptive.
Now, that sounds like I’m flippantly dismissing the entirely valid points that Sly Flourish, Alphastream, NewbieDM, and others have made. That’s not my aim here. What they’re mainly talking about is a narrower piece of the rules that define consequences around death – death by massive damage, death saving throws, and the great ease of saving someone with a healing word. Yes, the game does consciously make it more likely that a dying PC is saved – you have to fail 3 times, you succeed 55% of the time, and any healing or stabilizing effect wipes out all failures.
That’s a lot of potential levers to pull. Tomb of Annihilation proposes DC 15 death saving throws, which is an incredibly simple fix. Combining this with additional hit points at 1st level probably normalizes the game’s danger curve in a beneficial way. No amount of tweaking the DC does the trick if there’s a healer ready to go with healing word or spare the dying or whatever. I don’t recommend changing the ease of saving someone with healing word, though. Saving the dying is one of the most exciting gameplay moments for healers – don’t take that away from them.
Experimenting with making it more involved to wipe out failed death saves seems workable to me. For the smallest incremental change, maybe all failed death saves are erased when you’ve gone 1 minute without rolling a death save.
I’m also not at all sure that death from massive damage is worth trying to fix as a way that characters die. The fact that, by 5th level, you can get in way over your head and still not go from full health to instantly dead even from high-level monsters in a single hit – that’s a feature, not a bug. It keeps healers in the game. Full hit points is about as much of a high-confidence state as D&D grants. If you know you’ll have at least one chance to make a new decision and try to escape from almost anything, you don’t need to be quite so pathologically risk-averse – and D&D is really not about risk-averse people doing risk-averse things. (Ars Magica is one of the only major tabletop games that might be about that.)
Turning a bit away from the kinds of difficulty and danger that the inciting conversation was about, I want to suggest that a sense of difficulty and danger is more meaningful than leaning into high DCs or making characters more fragile. I think there are two Great Commandments when it comes to a sense of danger.
- Deliver a stinging defeat or pyrrhic victory early in the game. This establishes the sense of danger and consequence in your players’ minds. If you try to pivot later in the game to higher danger, they’ll feel betrayed.
- Threaten things that the PCs really could lose. This heightens the sense of dread because it makes them imagine their situation without that person, place, or thing.
Ask yourself whether you’re interested in running a pure power fantasy game, or if you want a game with a real sense of threat. These are both 100% valid! Neither approach is for everyone – this is the key decision step for player buy-in. The trust required for a “real threat” game is more complicated. You need the players to trust that a problem that is unassailable today will have a chink in its armor tomorrow. Even though they may lose something if they have to run away and live to fight another day, it won’t all be for nothing. If it seems that it’s all for nothing, then the story isn’t over yet.
The leading game-running advice for at least the last twenty or so years has been to make fights about something other than killing everyone on one side or the other. Do the heroes save the prisoner and escape from a force too large to fight? Does the demon destroy the altar? Is the healing spring befouled before the heroes get a draught from it that can restore the Fisher King? Yet we’re still not seeing published content engage with this tenet of best-practice encounter design. This is what I’m talking about with things the PCs could lose.
Even if you’re not chasing that “real threat” feeling, though, backing off on stakes once you’ve established them is one of the worst decisions you can make. There are ways to reveal that the stakes weren’t as dire as the PCs believed – a sudden moment of mercy, the work of a hidden ally, divine intervention – but those have to be handled with so much care. Otherwise the PCs dismiss the stakes of future conflicts, and increasingly expect to get away with treating even friendly NPCs like they are disposable. To put that another way, you’ve shown them that the world isn’t real, and they behave accordingly. This is a corrosive state of affairs even for a pure power fantasy game.