To begin to understand what solutions there are to this difficulty, first I want to explore exactly what I find unsatisfying with the way social encounters are ran in 5th Edition. We will return to the problems with exploration in later sections as a potential solution to these problems.
- Low Player Engagement. If the difference between success and failure is just a die roll, and nothing the player says or does effects the result of the social encounter, they are not engaged in the fiction of the encounter in such a way as to build personal investment. Nor are they challenged in any meaningful way. Whatever joy from success they get is the same joy someone gets from passively pulling a slot machine, but with much worse payout. Likewise, characters who are highly motivated and actively role playing in any given encounter may be frustrated when their impassioned acting becomes reduced to mere advantage, or worse, is treated as an annoyance or indulgent distraction before the more meaningful rolling of the dice.
- Low GM Support. The only leverage a GM has to reward good role playing is to lower the difficulty of a check (thus lowering the dramatic stakes!) or to bypass the check entirely. The former is an option commonly employed but never outlined in the rules of the game. The latter option can be occasionally interesting, as a way to reward player skill, but when done repeatedly can feel less like a player cleverly deducing the correct path based on context clues and more like a refusal to engage with the social pillar entirely. Moreover, not every GM is comfortable adjudicating complex NPCs motivations and desires without some kind of formalized support system put in place by the game, so such judgement calls can feel overly burdensome for people less used to inhabiting the stance of an actor or improviser.
- Arbitrariness. While there is a component of randomness to whether any given attack will hit, or whether an opponent will make their save, random chance is not the sole element of combat, and players who make good choices can reasonably count on victory (partially because the math is intentionally stacked in the favor). There is no such interplay if a player wants to persuade an an opponent to deescalate a combat. There is nothing but a binary choice, success and failure, separated by a somewhat nebulous DC. In some cases, a DC doesn't even feel appropriate, but players who have become accustomed to using social interactions in much the same way they use attacks to counter AC may feel frustrated when they are told that an NPC simply isn't vulnerable to being convinced or cajoled into a certain action. After all, in combat, a GM never declares a creature cannot be hit. This frustration mostly arises when players lack a sufficiently nuanced understanding of the NPC's motivation (whether this is because they had a lapse in attention or the GM did not effectively communicate their character traits). When players are frustrated or confused, the GM's choice to have certain approaches fail or succeed without involving rolls will seem less like an immersive experience and more like an arbitrary one. Choices that are fictionally sound in the abstract suddenly become little more than removals of player agency. Alternatively, a GM can decide that anything goes, which may be more fun in a mad cap way, but the players deceiving important NPCs without obvious falsehoods because of a high Charisma stat will over time erode the sense of reality.
- Binary Fail Conditions. While 5th Edition gives some lip service to the idea of nuanced failure states, the history of Dungeons & Dragons as a pass/fail system along with the fact that there are no concrete examples for adjudicating a mixed success mean that such protestations are just that, lip service. The most common roll players will make is an attack, which is pass fail. A saving throw may occasionally impose some consequence even on a success, but there is no obvious social equivalent to taking half damage. Powered by the Apocalypse games have long realized that fail states where nothing happens are death to interesting encounters, and have codified that failing an individual roll is not the same as failing at the task. A GM is merely given narrative license to introduce complications with each failure. A failed attempt to charm a person may end up a Pyrrhic victory, where the would be charmer finds the object of their attention to be more trouble than they're worth, attracting threats from all sides by other jealous suitors. This is codified as failing forward and while it is certainly an improvement on the most basic version of 5th Edition's conflict resolution, 5th Edition is not built on the bones of a story game. The idea that a skill check may technically succeed on some failed roll, but cause an unrelated bad outcome somewhere else in the narrative that is because of their failure but not fictionally grounded in their character's mistake may stretch the limit of some players suspension of disbelief. Some games attempt to make total and unequivocal failure interesting. Take Burning Wheel for example, whose guiding principle of let it ride makes the finality of failing central. When trying and failing to achieve a task is made core to progression, and when the possible approaches are nuanced and not abstracted, failure can be imbued with some meaning without introducing an outside force. This is more difficult in 5th Edition, in which failure is to be avoided at all costs, and even the loosest interpretation of milestone advancement only pushes characters forward on success. Moreover, it has a single social skill, Persuasion, which is buttressed by Deception (Persuasion with Lies) and Intimidation (Persuasion with Threats). Compare that to Burning Wheel's impressive list of social skills, which understands that not all social encounters are about convincing one individual to agree to do what you want in the most straightforward terms: begging, brutal intimidation, child-rearing, coarse persuasion, code of citadels, command, conspicuous, dignity of the wilderlands, drinking, etiquette, extortion, falsehood, foreign languages, haggling, howling, inconspicuous, instruction, interrogation, jargon, litany of fools, oratory, paean of deeds, persuasion, poisonous platitudes, preaching, ratiquette, religious diatribe, seduction, shrewd appraisal, silent fury, song of lordship, soothing platitudes, stentorious debate, suasion, troll etiquette, ugly truth, and voice of thunder. While it is easy to see how a skill like Command might fill the same narrative and mechanical roll as Intimidation in 5th Edition, it becomes less clear how one might model Poisonous Platitudes, where it is conceivable the goal is less about convincing an individual to do something and more about simply wounding them with your words. Likewise, a skill like Shrewd Appraisal begins to draw out an understanding that a skill like Insight could be a vital part of a more robust social encounter system, and not merely an emergency button players press to detect NPC lies in the place of using their own judgement. My personal favorite is Religious Diatribe, because the image of a character using Religion in the place of Persuasion to brow beat an individual into an action without ever explicitly asking them to do something, therefore preserving plausible deniability, is exactly the type of play I want to encourage with this system.
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/10091901/phantom_thread2.jpg) |
| A Still from Paul Thomas Anderson's Phantom Thread, a perfect deception of the fraught nature of conversation that does not have to lean on action scenes or combat to be made heightened and intense. |
Two Solutions: Combat or Exploration?
I posit two possible solutions. The
Dinner Party as Gladiator Pit approach models social encounters after combat while the
Dinner Party as Landmine models them after natural hazards experienced during exploration or the danger of complex traps. The advice 5th Edition actually provides for running social encounters (one of the most valuable parts of the Dungeon Master's Guide that every GM should familiarize themselves with) lays the groundwork for this experiment:
- Goal-First Interaction. 5th Edition only offers clear mechanical guidelines for Persuasion, centered entirely around what the party wants to make an NPC do. It makes the odd assumption that all attempts to be persuasive are truthful and non-threatening, making Deception and Intimidation de facto if not de jure sub-sets of broader persuasive activity. It does not leave open the option that NPCs would ever do something in response to being persuaded other than agree to the player character's terms. For instance, a character who tries to Intimidate a guard into revealing some secret about their master could not, under the assumptions provided in the DMG, simply flee from the players on a successful check. This means that the players decide what they believe the success condition of any social encounter would be, unlike combat and exploration where the purview of such things typically falls at the feet of the GM. While the Player's Handbook or Dungeon Master's Guide never explicitly says the GM should ask what the players their intention when making checks, it is so integral to the conflict resolution that it should become a formalized.
- Pseudo-Conditions & Hard Limits on Success. Creatures can be hostile, neutral, or friendly, and that these designations have material consequences before a dice is ever rolled. These statuses occupy a nebulous state, they have material effects like conditions but appear nowhere in the Player's Handbook. At first, they seem to be a piece of fiction first design in a system that is otherwise hyper-focused on mechanical detail, where things like being afraid or charmed mean very narrow and specific things, and even common sense statuses like unconsciousness and blindness must be relentlessly detailed. On deeper reading, it becomes clear these seemingly fiction-centered pseudo-conditions are just as systemic, but the effects they have are not explicitly detailed. When one examines the suggested DCs in the Dungeon Master's Guide for convincing hostile, neutral, or friendly NPCs to comply with the wishes of the player character's, it becomes clear that the designer's intend that a hostile NPC can never be convinced to accept any risk on the party's behalf, no matter how minor. Likewise, a friendly NPC can never be spurned in such a way that they act against the party's interests. This hard limit on the possible outcomes of success removes some of the fun of unexpected and emergent dynamics coming out of social interaction by narrowing the range of NPC responses to the characters. The designers may have assumed a GM was solving for this by constantly making judgement calls as to whether each roll interacting with the NPC shifts them from hostile to neutral or friendly, but this ad hoc shifting produces an uncomfortable asymmetry. It becomes assumed that players can only effect the game world by rolling dice, making ability checks with some perfunctory role playing attached, while the GM can at any moment reshuffle the emotional states of the characters on a whim. You may say, isn't that just what GM's do? Partially, yes, but much in the same way a GM can (but should not) change a monster's Hit Points in the middle of the fight, they should have some way of systemically modeling their characters that players can repeatedly and consistently interact with using their abilities. Without a framework it becomes impossible for a player to become skilled as social encounters, leaving it to numbers on a sheet.
If we were to adopt the Dinner Party as a Gladiator Pit, the solution to these aforementioned problems would be to make social interaction like combat. In the best case, this could be like Burning Wheel's the Duel of Wits, a clear system for mutual antagonism that meaningfully changes the internal motivations of those involved. The Duel of Wits assumes that a character's actions (though not their internal beliefs) can be forcibly decided after risking a social engagement and losing. While this nominally goes against 5th Edition's assumption that player agency is sacrosanct, the difference could likely be adjusted to. After all, the GM very rarely consults a player as to whether they're open to being attacked, losing Hit Points, or even dying, because we accept that drama comes from at least some token level of antagonism between the players and game world. The problem with the implementation of these mechanics is less that the type of play they encourage are alien to the landscape of 5th Edition, but that they are hamstrung by the limited tools a character has to systemically interact with the game world. For as much as its simplified skill list is a virtue, it can also be a vice, because complexity brings with it nuance. If even the absurdly well supported combat pillar of 5th Edition can feel like an unpleasant distraction from the narrative play many players have now come to expect, mimicking combat with even fewer meaningful choices or ways to differentiate player characters would inevitably replicate the feeling of sameness that can be birthed from repetitive play. It is also not as intuitive as one might believe at first blush, even if on its face it is copying a system people know well. While long time players have sat with combat's assumptions for years, slowly understanding its idiosyncrasies on their own terms, they would have no conception how many Hit Dice an argument should have. Hit Points are already a confusing abstraction when applied to bodily health, despite decades of TTRPG and video game history elevating them to the status of a sacred cow. A more simplified rock-papers-scissors system in the vein of Burning Wheel would be easier to internalize despite not hewing identically to 5th Edition's mechanical palette (I.E., Deception beats Persuasion, Persuasion beats Intimidation, and Intimidation beats Deception) but these kind of hard lines feel like game conceits. If given the choice between engaging with combat or social encounters, it would be very difficult to imagine players choosing the system with less mechanical support, no matter how well implemented. In such a malformed state, players would inevitably favor the version of two similar systems of antagonism that has more robust support, where it is clear by virtue of the things that appear on their character sheet how they may contribute, preferring to keep the distraction of social encounters contained within a single rol1. They may even be right to want such a thing! The challenge is not merely to make social encounters resemble combat because combat is fun, its to find what can be fun about social encounters are their own, distinct entity, and provide mechanical support for that play!
This is why I ultimately prefer what I call the Dinner Party as a Landmine approach. To begin to outline its usefulness, let us return to the other undeserved pillar of play in 5th Edition, exploration. While certainly deserving of its own post, at the very basic level, the difference between good and bad encounters during exploration is whether they feel perfunctory or meaningful. Whether in the wilderness or in the dungeon, whether facing a natural hazard or constructed trap, challenges imposed by the GM when they fit well into the overall loop of play. The temporary addition of resource management for one narrative arc or travel will fall flat in a game that doesn't place value on encumbrance anywhere else. Likewise, random encounters fall flat when player's feel like they are just draining resources or wasting time, guaranteed to result in either unlikely death or minor inconvenience forgotten after one night's rest, with no in between. In fiction outside of D&D, we see wilderness survival and exploration as being interesting when it is not only a puzzle to solve, but a threat with a sense of permanency. The madness imposed by isolation, the terror at the sublime fury of nature, the way an expedition may turn in on itself when supplies run short. All of these come after the puzzle-solving element of nature, as failure conditions that emerge when an individual or group is unable to impose their will on an indifferent world. A trap in an Indiana Jones movie or a bomb defusing scene in an action movie can often feel different from on in a tabletop game because the individual's skill of trap disarmament is less emphasized than the aftermath of the disaster. What does Indiana Jones do after the boulder is rolling and darts are flying? What forces avail themselves to prevent the disarming of the bomb? What would happen if it did go off? The moment when you begin to think the protagonist failing to disarm the bomb is a possibility is the moment when the scene fully grasps you in the drama. All of this could be said of the drama of social interaction, because the social is a sphere as fraught as any romantic or idealist depiction of the wilderness, or the most tense scenes of action and adventure. A dinner party full of people with hidden agendas, personal flaws, complex motivations, is like a landmine that needs to be gingerly defused, not a brawl. Mistakes can cause lasting consequences that are interesting, that goes beyond 5th Edition's way of modeling of failure, which is to simply not give players what they want when they roll poorly. This can be made more interesting by a good improvisational GM, but we shouldn't have to rely on that.
To summarize, a good social system should:
- Engage the player by asking for more than a dice roll
- Offer a way for GM's to adjudicate success and failure more clearly and fairly
- Offer clear ways to determine NPCs disposition and how it may change
- Offer options for graduated success, and options to make different options feel unique
- Give openings for unexpected and improbable results
- Avoid replicating the slog of combat and provide a novel type of game interaction.
- Make failure interesting, avoid temporary frustrations or setbacks
The Final Product: NPCs and Social Hazards
The first draft of this new system contained
Personality Traits that could be given to any NPC significant enough to be the likely target of social interaction. It assumed that a basic but recurring character (say a fence for stolen items) may have a single defining trait, while a major antagonist or ally may have up to three that could be discovered in time through playing the game. These quirks served the same role as triggers for a trap. When players interacted with them, they modified the difficulty, danger, and effectiveness of social interaction. In the case of less developed NPCs, a player might surmise that a character was vulnerable to flattery or resistant to threats of violence, modifying only the ease of success. For a more developed NPC, they may be vulnerable to flattery, but for the modified ease, they may find that the character is also more volatile, making the rare failures to flatter catastrophic social failures rather than mere setbacks. Likewise, a character who was resistant to threats of violence who, was moved by an unusually successful Intimidation check, may find themselves even more pliant than the average NPC. This produces a tension by emphasizing certain skills over others regardless of what players may be specialized in, and also introduces a risk. A player would have to choose whether, knowing what they know about a character, pushing a particular button would give them leverage or blow up the whole negotiation. A character could be
Paranoid:
Paranoid. All Deception checks against [insert character here] have disadvantage because of their paranoia. A player character who discovers the root cause of [insert character here's] paranoia can use that information to become sympathetic, no longer suffering disadvantage. Any revelation that the party knows the secret or has attempt to use the secret as leverage immediately causes the NPC to become hostile, but gives players advantage on Intimidation checks.
This provided a solid framework but created a few problems of its own. For one, it only provided one dimension to model a character, their internal personality. To make this system broader, it would have to take into account the social context that extends beyond individuals. For instance, two likeminded and well dispositioned people may still find themselves in conflict over things like a difference in class or ethnic background, given a conservative political climate. Likewise, other fictional concerns, like the secrets held by some shadowy underground organization, might bind all of its members with shared concerns despite their many different dispositions. Broadly, these contextual concerns can be split into two major types: Social Traits and Organizational Traits, making an NPC distinguishable by the context of their culture or society, and by any factional interests they may have. It is intended that these tools provide GMs more material to effectively build their worlds and generate interest about the fiction of their game, as players become more and more grounded in the reality of the setting based on the mechanics of the game.
These additions may make it easier for a GM to communicate more complex NPCs, especially for GMs who feel less comfortable in the actor stance, who find it difficult or prefer not to embody their roles but instead deliver information about their characters in a more abstracted way. While it may be difficult for such a GM to distinguish between a hyper-paranoid NPC or a merely aggressive NPC, giving them options to use the fiction of the game world will allow the player's to do some of the heavy lifting for them, instead of placing all aspects of this system on how skilled a thespian the GM is. It also broadens the potential pitfalls a party can fall into, since its not just their skill choice that can get them in trouble, but their backgrounds, classes, and other ephemera decided at the point of character creation and discovered during play.
 |
| Still from neo-noir classic Chinatown which I am definitely not ripping off for the sake of this illustrating example. |
Lets take the example of the iconic noir detective to draft an NPC with these systems. If the characters wanted to hire the services of a private eye like the inimitable Heracles Flint, they may be confronted with the following idiosyncrasies that make the encounter more fraught (remember, you wouldn't need to use one of each trait type, a character could simply have three personality traits, but this example is only to show off the potential of these systems).
Personality Trait. Heracles Flint is full of bluster. All Charisma (Intimidation) checks are made with disadvantage. On a successful check, Heracles Flint becomes hostile instead of neutral, but will do as he's told so long as he's compensated, dragging his feet instead of outright disobeying. Getting him to disclose what he's discovered in full will require a successful Charisma (Persuasion) check, otherwise, he distract the party with vagaries and half-truths, only technically complying with requests, attempting to waste the party's time until it is too late.
Organizational Trait. Heracles Flint is a private eye, and as such, is known to the local Military Police of La Colina. Some consider him their equal based on past dealings, but often, his reputation will precede him in negative ways. Any attempt to associate with him will likely result in the police taking note of your patronage as well and attempting to thwart him by getting to you directly, especially from those who may otherwise by on the dole of corporate security.
Social Trait. Heracles Flint believes the world is fundamentally unjust, but because of his jaded outlook, he seeks to protect others from their "sheltered" upbringing. Players who respond to Flint's warnings with innocence or naivete (genuine or faked) can improve their status with Heracles Flint from hostile to neutral. If they are the actual victim of the injustice, and not some would be crusader, they can further change Flint's status from hostile to neutral or from neutral to friendly.
Social Encounters Coda: Making it Quick Easy
It is totally conceivable that the above information could be easily fit into an NPC card, much like the cards used to outline a monster's abilities in combat. It would make laying out the stats of a non-combat NPC more meaningful, given the higher likelihood that their stats may actually end up mattering in a world where they and the Party do not come to blows. The only question remains are, what target numbers should a GM be keeping in mind for NPCs as far as social encounters are concerned? I believe that an NPC could have three stats listed by default: Passive Insight, Investigation, and Persuasion, to cover a broad range of the normally applicable rolls, and other skills that may be relevant based on their proficiencies. For instance, a notable cleric of some divine order may have their Passive Religion laid out by default to remind the GM that they should be encouraging their players to think creatively with their skills, and telegraph that theology is as legitimate a social pursuit as other Charisma skills.
When a player rolls Deception, Intimidation, or Performance they roll against either Passive Insight or Passive Investigation, dependent on whether their lies, threats, or acting is appealing to emotion or logic. This makes wise and intelligence NPCs resistant to these approaches by default.
When a player rolls Persuasion they roll against Passive Persuasion, making charismatic characters harder to convince through raw charm or force of will.
When a player rolls another relevant skill (usually Intelligence based) to make a case based on their expertise, they roll against either Investigation (if the NPC is a non expert) or the same skill (if the NPC is also proficient). This assumes that the player is speaking truthfully, otherwise the relevant skill would be deception, regardless of whether they were incorporating their knowledge of other subjects.
Then we can add variation based on the hostile, neutral, and friendly pseudo-conditions:
A player has advantage on Ability checks made in social interactions with friendly NPCs, and hostile NPCs have advantage on Ability checks made in social interactions with players (effectively raising their Passive DCs by 5).
