Friday, June 24, 2016

Bounded Accuracy

[Originally posted at but no longer available.]

Conventional D&D wisdom tells us that the maxim "the numbers go up" is an inherent part of the class and level progression in D&D. While that might be true, in the next iteration of the game we're experimenting with something we call the bounded accuracy system.

The basic premise behind the bounded accuracy system is simple: we make no assumptions on the DM's side of the game that the player's attack and spell accuracy, or their defenses, increase as a result of gaining levels. Instead, we represent the difference in characters of various levels primarily through their hit points, the amount of damage they deal, and the various new abilities they have gained. Characters can fight tougher monsters not because they can finally hit them, but because their damage is sufficient to take a significant chunk out of the monster's hit points; likewise, the character can now stand up to a few hits from that monster without being killed easily, thanks to the character's increased hit points. Furthermore, gaining levels grants the characters new capabilities, which go much farther toward making your character feel different than simple numerical increases.

Now, note that I said that we make no assumptions on the DM's side of the game about increased accuracy and defenses. This does not mean that the players do not gain bonuses to accuracy and defenses. It does mean, however, that we do not need to make sure that characters advance on a set schedule, and we can let each class advance at its own appropriate pace. Thus, wizards don't have to gain a +10 bonus to weapon attack rolls just for reaching a higher level in order to keep participating; if wizards never gain an accuracy bonus, they can still contribute just fine to the ongoing play experience.

This extends beyond simple attacks and damage. We also make the same assumptions about character ability modifiers and skill bonuses. Thus, our expected DCs do not scale automatically with level, and instead a DC is left to represent the fixed value of the difficulty of some task, not the difficulty of the task relative to level.

We think the bounded accuracy system is good for the game for a number of different reasons, including the following:

Getting better at something means actually getting better at something. Since target numbers (DCs for checks, AC, and so on) and monster accuracy don't scale with level, gaining a +1 bonus means you are actually 5% better at succeeding at that task, not simply hitting some basic competence level. When a fighter gets a +1 increase to his or her attack bonus, it means he or she hits monsters across the board 5% more often. This means that characters, as they gain levels, see a tangible increase in their competence, not just in being able to accomplish more amazing things, but also in how often they succeed at tasks they perform regularly.

Nonspecialized characters can more easily participate in many scenes. While it's true that increases in accuracy are real and tangible, it also means that characters can achieve a basic level of competence just through how players assign their ability bonuses. Although a character who gains a +6 bonus to checks made to hide might do so with incredible ease, the character with only a naked ability bonus still has a chance to participate. We want to use the system to make it so that specialized characters find tasks increasingly trivial, while other characters can still make attempts without feeling they are wasting their time.

The DM's monster roster expands, never contracts. Although low-level characters probably don't stack up well against higher-level monsters, thanks to the high hit points and high damage numbers of those monsters, as the characters gain levels, the lower-level monsters continue to be useful to the DM, just in greater numbers. While we might fight only four goblins at a time at 1st level, we might take on twelve of them at 5th level without breaking a sweat. Since the monsters don't lose the ability to hit the player characters—instead they take out a smaller percentage chunk of the characters' hit points—the DM can continue to increase the number of monsters instead of needing to design or find whole new monsters. Thus, the repertoire of monsters available for DMs to use in an adventure only increases over time, as new monsters become acceptable challenges and old monsters simply need to have their quantity increased.

Bounded accuracy makes it easier to DM and easier to adjudicate improvised scenes. After a short period of DMing, DMs should gain a clear sense of how to assign DCs to various tasks. If the DM knows that for most characters a DC of 15 is a mildly difficult check, then the DM starts to associate DC values with in-world difficulties. Thus, when it comes time to improvise, a link has been created between the difficulty of the challenge in the world (balancing as you run across this rickety bridge is pretty tough due to the breaking planks, especially if you're not a nimble character) and the target number. Since those target numbers don't change, the longer a DM runs his or her game, the easier it is going to be to set quick target numbers, improvise monster attack bonuses and AC, or determine just what kind of bonus a skilled NPC has to a particular check. The DM's understanding of how difficult tasks are ceases to be a moving target under a bounded accuracy system.

It opens up new possibilities of encounter and adventure design. A 1st-level character might not fight the black dragon plaguing the town in a face-to-face fight and expect to survive. But if they rally the town to their side, outfit the guards with bows and arrows, and whittle the dragon down with dozens of attacks instead of only four or five, the possibilities grow. With the bounded accuracy system, lower-level creatures banding together can erode a higher-level creature's hit points, which cuts both ways; now, fights involving hordes of orcs against the higher-level party can be threatening using only the basic orc stat block, and the city militia can still battle against the fire giants rampaging at the gates without having to inflate the statistics of the city guards to make that possible.

It is easier for players and DMs to understand the relative strength and difficulty of things. Under the bounded accuracy system, a DM can describe a hobgoblin wearing chainmail, and, no matter what the level of the characters, a player can reasonably guess that the hobgoblin's AC is around 15; the description of the world matches up to mechanical expectations, and eventually players will see chainmail, or leather armor, or plate mail in game and have an instinctive response to how tough things are. Likewise, a DM knows that he or she can reasonably expect players to understand the difficulty of things based purely on their in-world description, and so the DM can focus more on the details of the world rather than on setting player expectations.

It's good for verisimilitude. The bounded accuracy system lets us perpetually associate difficulty numbers with certain tasks based on what they are in the world, without the need to constantly escalate the story behind those tasks. For example, we can say that breaking down an iron-banded wooden door is a DC 17 check, and that can live in the game no matter what level the players are. There's no need to constantly escalate the in-world descriptions to match a growing DC; an iron-banded door is just as tough to break down at 20th level as it was at 1st, and it might still be a challenge for a party consisting of heroes without great Strength scores. There's no need to make it a solid adamantine door encrusted with ancient runes just to make it a moderate challenge for the high-level characters. Instead, we let that adamantine door encrusted with ancient runes have its own high DC as a reflection of its difficulty in the world. If players have the means of breaking down the super difficult adamantine door, it's because they pursued player options that make that so, and it is not simply a side effect of continuing to adventure.

This feeds in with the earlier point about DMs and players understanding the relative strengths and weaknesses of things, since it not only makes it easier to understand play expectations, but it also ties those expectations very firmly to what those things are in the world. Now, we want to avoid situations where DMs feel bound by the numbers. ("Hey," says the player, "you said it was an iron-bound wooden door and I rolled a 17, what do you mean I didn't break it down?") We hope to do that by making sure we focus more on teaching DMs how to determine DCs and other numbers, and letting them adjust descriptions and difficulties based on their needs.

-Rodney Thompson, D&D designer

If I esteem mankind to be in error, shall I bear them down? No. I will lift them up, and in their own way too, if I cannot persuade them my way is better; and I will not seek to compel any man to believe as I do, only by the force of reasoning, for truth will cut its own way.

I could not love thee, dear, so much,
Loved I not Honor more.

Wednesday, June 22, 2016

On X-Projects

On X-projects:

"The typical X project focused on a needed technology. Although the technique is applicable to many areas of technology, I'll stay with aerospace for the moment. One designs the best ship possible given existing technology. There are to be few to no stretches or reaches: we are not looking for new technology, we are looking to see what the best we have can do – and thereby identify what's needed next.

"The ship is built. Typically there will be three vehicles (tail numbers in the jargon). The first is flown to find out its capabilities. Then those limits are tested, and tested again. Frequently tail number One is destroyed in the test process, although that's not inevitable. Using what was learned from One, Two is modified and flown to its limits, and kept flying until there is no more to learn. Number Three makes a few token flights and goes to the Smithsonian."

It works in software too. "Build one to throw away." This is why the planning phase between sprints is important.


If I esteem mankind to be in error, shall I bear them down? No. I will lift them up, and in their own way too, if I cannot persuade them my way is better; and I will not seek to compel any man to believe as I do, only by the force of reasoning, for truth will cut its own way.

I could not love thee, dear, so much,
Loved I not Honor more.