Courtney Cambell, of the Hack & Slash blog, recently released a supplement for handling NPCs and social interactions, DM3 On the Non-Player Character. It contains many useful tools. Here is a discussion of some of them and a consideration of the product as a whole.
The form factor is a 62 page saddle-stapled paperback booklet. This is a good size, due to its convenience and portability. Using this format also means that the PDF renders nicely on a tablet without zooming. The book is divided into sections on the core social encounter system, persistent relationships (“agents”), creating NPCs (with numerous tables), and a social combat system built around subdual damage. The focus is on game systems that interact with player choices, such as what NPCs want, what they can offer PCs, and what specific interactions trigger various results. Less emphasis is placed on general motivation, though there are some tools for integrating personality traits into the design of NPCs.
A quick note on the price. Courtney has decided to set his price high relative to current market rates, around $30 for the print version and $20 for the PDF. I suspect this will cause perturbation in some individuals. Niche RPG supplements are not necessities, however, so I don’t think Courtney has any obligation to price competitively. Personally, I think supporting thoughtful hobby publishing is worth paying the premium. I’ve also gotten lots of use out of his other supplements, which are available for free, on treasure and trap design, not to mention his blog, and this feels like a nice way to give back and also get something useful in return.
The core of the social interaction system should be familiar to players of traditional D&D. The referee makes a 2d6 roll, modified by PC charisma and circumstances, to determine NPC reactions. The DM3 version of this system has PCs continue to make 2d6 rolls for every social action attempted. The initial roll determines the basic disposition (which imposes a modifier to the following social action rolls) and the number of social actions the PCs may take before the NPC ends the encounter. This differs from the system in the original Basic line: “Don’t roll more than three times. If by the third roll the monster hasn’t achieved a roll of 10 or better, it will decide to attack or leave” (Rules Cyclopedia, page 93).
Each further social action is resolved as one of several options. There are 18 different “moves” provided, including options such as converse, question, and threaten. All of these have game weight. Some of the options provided are surprisingly specific, such as drink, which indicates an offer of alcohol. Players can either state their move directly in game terms (“I intimidate the NPC”) or describe actions and rely on the referee to interpret which move is attempted. Each move includes a target number for the 2d6 roll, potential modifiers, potential results, and so forth. The number of options is a bit overwhelming, and the differences between the various moves also adds to the complexity. Luckily, there is a nice one-page summary table near the back of the book. Any referee that wanted to use this system as written would need a copy of this chart at hand, at least for a while. There are also four possible “stances” that PCs can take which may interact with NPC personalities. The stances are neutral, hostile, friendly, and obsequious.
I haven’t tried to run an encounter using all the options, but I suspect that I would want to simplify them schematically somewhat (perhaps into friendly, neutral, and hostile acts, sort of like the stance categories) and generalize the target numbers so that I wouldn’t need to constantly reference a table. The specifics do create some interesting possibilities though, such as the ability to convert NPCs using the (somewhat awkwardly named) pray action (which plugs into the bond system, discussed below). So I think there would be some value in using the whole, baroque mechanism. Obviously, some parts of the system could be combined with improvisation, so this is not an either/or.
I do think that using the codified system to its full potential would require some practice. It is important to note that it is not necessary for players to master this list of moves in order to engage with NPCs, but if they don’t the referee will need to be able to translate diegetic PC actions into social moves. Might this introduce an element of system mastery into this part of the game? That is, would a player that studied the moves carefully and understood the underlying system more comprehensively have an advantage over a casual player? Perhaps, though a reasonable and attentive referee should be able to mediate between the player and the rules well enough to avoid that issue (in the same way that a referee might need to know when to apply the rules for a contested ability check or bull rush during combat even if the player does not invoke the rules directly). This issue probably deserves more attention, especially given Courtney’s position against “fiction first” (that is, diegetic or game world focused) player interfaces, but that is a subject for another post.
DM3 distinguishes between NPCs that are are strangers and NPCs that are known, which are called “agents.” This terminology is not my favorite (“agent” has so many other meanings), but I suppose you have to call them something. Persistent NPC state relative to PCs is measured by a bond number, which is the core of the persistent relationship, mechanically. This is a brilliant system, and is probably my favorite part of the overall framework. Every interaction has a chance of converting a normal NPC into an agent, or increasing the bond of an NPC that is already an agent. Bonds can represent anything from employee to rival, and there are several custom social action moves that work specifically with bonds, including gift and seduce (which is used for any conscious attempt to improve a bond, not just sexual seduction). I could easily imagine collecting bonds on a PC’s character sheet, which would be an interesting way of letting a character develop mechanically outside of the standard level/power and gear tracks that D&D characters usually follow. I woud probably add a house rule to periodically decrease bonds with neglected friends (you never call, you never write!).
Creating an NPC requires noting down a few stats, but not the stats you might expect (strength, dexterity, HP, etc). Instead, they are a few numbers that interact with the social actions and stances (and by default, of course they are all zero, so you only need to specify the interesting differences). This format looks very friendly to simple stat blocks, especially compared to the walls of text that are the usual presentation of NPC motivations. This method for recording the important (gameable) parts of NPCs is quite elegant. It also includes locks for keeping track of how a given NPC reacts to specific types of interaction. I could see using this record format almost unmodified, and it might be adaptable to other, abstract entities such as towns, organizations, or other aggregate bodies. NPC locks probably deserve clues as well, just like traps or hidden features in a dungeon, to help make sure that players can gain access to the interesting potentialities available, but the most efficient format for such clues probably requires more experimentation. A huge volume of tabular data is also provided to serve as imagination fuel for quickly creating a memorable NPC.
A social combat system is included that makes use of hit points and subdual damage. This doesn’t really fit into the way I usually run games, but I can see how it might be useful to others. It requires many calculations that look like 10 + Wisdom OR Morale + 1/2 hit die (+4 if Animal Intelligence). You also need to track at least five different kinds of damage separately (damage types include fear and confusion) which each have different effects and effect thresholds. It looks like it would work if you don’t mind shuffling the numbers (at base, it’s still pretty much just subdual damage), but I think it’s more complexity than I want to handle in play. This subsystem is not necessary to the other systems presented in DM3, however.
There are a few ad hoc rules that could probably be better handled with a core mechanic. For example, monsters (that is, unintelligible NPCs) that get the “freeze” result have a 20% chance to attack and an 80% chance to flee. Am I ever going to remember that during play? Nope. Am I going to spend the time to look it up during play? Highly unlikely. The few fiddly bits like this can be easily ignored though, without affecting the basic integrity of the system. There are also some references to other unexplained house rules that don’t really detract from the work as a whole, but do stick out. For example, what are alchemists, jesters, and swashbucklers? None of those are standard classes, but they are mentioned alongside fighters, magic-users and other, more traditional classes. The fivefold encounter area categorization scheme from DM2 is also referenced (empty, treasure, special, trick, and trap) without much explanation. Some extra text to explain these references would be helpful, or even just a footnote mentioning the external work. Otherwise, it feels a bit like an excerpt from a larger work.
The NPC modifiers to stances make me think that something like a general “temperament” score would be useful for monsters or NPCs. I don’t think I have ever seen anything like such a stat in a game before. It is not exactly the same thing as morale, which is more about confidence. My first attempt at such a stat would probably just be a modifier to the initial reaction roll. For example, something like a wolverine (or other nasty critter), might have something like a default temperament of -4, making it much more difficult for the initial reaction to be positive. Why isn’t there something like this already? It seems so obvious. Maybe this is something that is expected to be a situational modifier, but to me it seems like such a thing should depend on the nature of the creature or NPC (some creatures or people are just naturally prickly).
DM3 contains lots of food for thought and a tightly focused, comprehensive system for resolving social interactions impartially. I am definitely going to use some form of the bond system, though I likely will modify it to be a scale that stretches from enmity to friendship rather than multiple possible tracks each with a single magnitude. The NPC locks are an excellent way to include NPC depth in a way that is instantly relevant to what PCs do in the game. Even if you don’t agree with Courtney’s dictum that NPCs must be designed, slightly simplified versions of some of these rules could greatly assist in running impartial NPC encounters based on player skill. Just considering the issues covered systematically is quite valuable, as there are very few other discussions of the higher level design issues involved in social encounters.