Bounded Accuracy
I’m obviously pretty far behind on posting my thoughts about Legends & Lore, which I should have seen coming what with the stupendously successful Kickstarter, big work announcement, and my wedding. Well, it’s time to play catch-up, so let’s take a little jaunt back to Rodney Thompson’s post from over a month ago on Bounded Accuracy.
I found Rodney’s entire post very interesting mostly because it described something I really like in terms that make it sound pretty obtuse.
Rodney’s post describes this entire system as a patch on how certain things are handled—a way of making something that was broken work. It’s all over the place as far as what this is supposed to do, which is probably because of Next’s focus on pleasing everyone: one point talks about how this makes for better scenes, as if D&D was a drama to be played out, while another talks about verisimilitude, and some points are entirely procedural.
So let’s take a step back and look at what target numbers for rolls (and therefore the bonuses to those rolls) mean—what they say about the world of the game—in different systems.
First let’s look at World of Warcraft and Dungeons and Dragons 4th Edition. I don’t really buy that those two are all that related, but they do use target numbers in a similar way. In both games target numbers rise with level to effectively mark content (encounters, instances, whatever) as only available to certain levels. In WoW a level 60 instance has numbers (of attack, damage, etc) so that only characters of a certain level with certain gear have a chance. In 4th edition the same thing happens. A character’s specific chance of succeding can be a bit higher or lower depending on player choices but the overriding factor is always character level vs. opposition level.
This ends up being an interesting statement about the world itself—the world is essentially divided into levels of ability in a pretty stark way. A 1st level character cannot succeed at forging a trail through the Woods of Despair. A 10th level character has a chance to succeed. A 20th level character can do it blindfolded. It implies a highly ordered and grouped world.
Essentially this is the Red Queen Hypothesis all over again1. 1. How cool is it that I’ve got to link to the Red Queen Hypothesis twice from articles on game design? Really cool. Very very cool indeed. Leveling up means facing threats that are relatively as powerful as you are. You go from having a 75% chance of hitting goblins to having a 75% chance of hitting orcs and so on.
Which is actually kind of similar to how early editions of D&D handled what we might call skills now. The Thief has a set chance of opening locks, no matter the lock, based on level. For the scope of Moldvay D&D this is a fantastic answer, but with the bigger scales that D&D has taken on (from lowly adventurers to plane-hopping demigods) it needs a little more information.
Which brings us around to Apocalypse World (and Dungeon World) where, like Moldvay, the chance of success is set but the game explicitly handles questions of scale by saying that the rules only apply in certain situations. For example, Dungeon World rule for picking a lock only applies when your character takes action to open a lock. The Moldvay rule can be read as allowing a character that chance no matter what (though that probably isn’t the intention), so that a 1st level thief has the same chance to pick the lock on the door to heaven as the lock on the door to the privy. Dungeon World says that you only roll to unlock a door if you take action to unlock it (i.e. action that could conceivably open the lock) so while the mathematical difficulty is set the overall difficulty—how hard it is to get the door open—still scales. Opening the picking the lock on the door to heaven probably takes some kind of divine (or infernal) lockpicks, access to the door itself, etc.
Vincent Baker calls this fictional positioning, and it takes us full circle to “bounded accuracy.” To me what “bounded accuracy” is actually saying, behind all the sales pitch, is that DCs reflect some aspect of the fictional world, and that player character skill also reflects some aspect of the character beyond abstract level.
That sounds a lot like d20, but with a key difference: in d20 player character skill still primarily reflects level. The DCs in d20 were always talked about as reflecting some aspect of the fictional world, but in my experience they more often were dictated by player level (i.e. “The Rogue has +8 to Open Lock, so the locks around here better be around DC 18… whatever that means.”).
Ultimately this leads to the same kind of fictional positioning as the AW system, just coming at it from another direction. Once higher DCs are not mandated by mathematical requirements they can freely reflect the fictional situation and the way that situation changes.
The interesting interplay here is with HP, which we’ve already seen (in the playtest docs) automatically increases with level. In the case of the Fighter that increase is pretty considerable: 6 HP per level. This creates kind of an odd dichotomy that I hope they’ll address: the ability to hit/defend has been freed from the constant escalation of levels, but the ability to take/deal damage still needs to scale. For all the talk of monsters still being threatening at all levels, if the Fighter’s HP increase continues at 5th level they’ll be able to take twice as much damage—a goblin doing 5 points of damage has stopped really meaning anything.
I try to mostly stick to discussions of other people’s games in these posts, but this is one place that I’ve got to speak from experience: we hit this exact same spot with Dungeon World. Everything but damage and HP has set numbers with the ability to invoke that rule dependent on the fiction, which has pretty much the same effect of DCs set by the fiction and non-automatic skill advancement. That worked great until HP and damage started scaling, and then it just broke down. There was a disconnect between one part of the rules and the rest: fiction-based rolls seemed to say that everything was fundamentally on the same scale, but level-based damage and HP implied that just being higher level put you in a different bracket of existence. Our solution was to make HP and damage fiction-dependent as well, instead of having them go up automatically with level. I wonder how Wizards will face the problem.







I think the idea in D&D Next that creatures are threatening at all levels is handled by the to-hit rolls rather than the damage. There were points, especially in 3rd Edition, where a creature couldn’t hope to hit the players.
4th Edition first attempted to fix this by including minions which could hit but with less damage and 1 hp each.
Next seeks to fix this by removing the need to create special minions and have goblins able to hit level 10 players so they can serve that role.
Great summary of the problem. I think Wizards will probably just smash in more goblins as you level up to “solve” this problem, creating a whole new long combat problem.
Huge swarms of enemies are a common trope of adventure stories. I’ve been reading Edgar Rice Burroughs Barsoom series where this is common. D&D isn’t equipped to handle this. A game where the fiction leads the rules could by using the same technique a novelist does, describing the action in a type of fast forward with key moments highlighted. The heroes could mow down faceless monsters until a tougher one appears requiring a roll.