On creative agenda and GNS

GNS theory is a hobbyist theory of roleplaying (as opposed to academic theory), which claims to classify creative contributions when roleplaying. Big model is a generalization of it.

For an introduction to the theory, see:

There are also several documents one should not use as introductions or summaries of the theory. A main figure in developing the theory was Ron Edwards, whose essays can be found on the Forge website: http://www.indie-rpgs.com/articles/. The articles of Edwards are, at the moment, historical documents, and do not explain the most recent thinking about the theory. As such, relying on them is not a good idea. More recent thinking of Edwards on these types of issues can be found on his Adept play website, but for most of the late Big model thoughts there is no good source. Adept play: http://adeptplay.com/seminar-hearts-minds/discuss-phenomenology

A useful critique of GNS is given by Zak Smith, https://dndwithpornstars.blogspot.dk/2018/03/the-impossibility-of-fastball-special.html. Zak reads an old GNS essay and comes to the conclusion that GNS categories, at least as defined there, are useless. On the way he observes that while writing the text, Ron Edwards did not really understand challenge-focused play that uses simulatory or complex rules, and also analyzed games in a very strange manner.

What follows is my taken on the concept of creative agenda and GNS categories; including how they are and are not useful. My definitions of the concepts in question are my own, but as far as I know (and as verified by occasional discussions) they are quite compatible with the definitions used by others who are familiar with the Big model.

Creative agenda

  1. Social interaction with non-awful people is often interesting. A related term in Big model is “social contract”. Its point is that all roleplaying, fundamentally, requires social interaction.
  2. Roleplaying (with non-awful people) is even more interesting. The process of roleplaying – sharing the fiction, contributing to it, and having your contributions validated from moment to moment – is called “exploration” in Big model. Here, validation simply means that others react to your contributions and build on them, as appropriate.
  3. Play might have more purpose to it in addition to the baseline of exploration. If the group members socially reward the creative contributions of each other (by verbal and non-verbal means), and do this consistently, then the theory calls the play of the group is “coherent”. A “creative agenda” is the patterns of social recognition and shared understanding of what good play means, for this given group, when they are actually playing.

An example of play with a creative agenda: When I am running or playing OSR games, typically everyone around the table understands that tactical and strategic decisions and creative solutions to problems are interesting, and being good at this is appreciated, while making poor choices is lamented.

Another example: A game with involved and interesting combat rules. Group creates a charged situation with characters capable of violence, and players pick and choose their characters from among these. Play consists of a violent drama between the characters (player and non-player alike), with tactical combats as appropriate. Everyone enjoys the alliances, betrayals and questions of trust and loyalty, as well as the tactics of the characters clashing.

An example without a creative agenda: Players create characters, but the group does not really know how roleplaying works (they just started based on reading some book and hearing rumours), so the GM says that they are walking on a forest road and there is a meat-eating plant blocking it, so some dice are rolled and it is killed, and then they come to a city and meet some slavers, and so on. The group might very well find a creative focus in all of this, but it is also possible they are fumbling around trying to figure out how to find some direction. Many of my first roleplaying sessions as a kid were like this; later, I figured out how to roleplay with purpose. Growing up also helped.

Another example without creative agenda: Group is there mostly just to hang out, tell jokes, eat and drink. The game provides a structure and inspiration for the jokes and the stories, as well as an excuse for meeting once a month. They might also enjoy the process of roleplaying, but it is not the focus of the activity (for more than one of them.)


According to GNS theory and Big model, there are three families of creative agendas (that have been observed thus far).

  • Story now (narrativism) is the creative agenda where group appreciates decisions with human interest – questions of loyalty, morality, faith, love, etc. The basic unit of story now play is a character in a situation where they need to select among several values; player has the character make the choice; the group plays to see consequences of the choice. This could be very aggressive and fast, or happen slowly in a more relaxed gameplay.
  • Step on up (gamism) is the creative agenda of showing one’s skills and daring. The basic unit of play is making a choice (such choosing to take or not take a risk), and playing out the consequences of that choice: does it take one closer to accomplishing one’s goals (winning), or not?
  • Right to dream (simulationism, constructive denial) is the creative agenda of investigating and experiencing some part of the gameplay – setting, character, rules system, or even situation. The basic unit of play is playing through something new, thus learning something about the focus.

USe of GNS

The GNS categories, and especially examples within each that can be found, for example, at the Forge articles named according to them, show the diversity of roleplaying. They also provide a rough vocabulary and framework for understanding the creative impulses in roleplaying.

That is, I do not see plenty of value in the GNS categories themselves, once one has figured out that roleplaying games can be creative in different ways, and gotten exposed to a representative sample (by experience or theoretical reading, or combination thereof).

Problems with GNS

In addition to providing a language and a conceptual framework, GNS theory also makes some claims. These are mostly of academic nature, in the sense that they are natural questions to ask (once provided with some kind of classification), but also in the negative sense that their applicability to issues of analysis and design of games is far from obvious.

Exclusiveness of GNS categories, or incoherence between categories

GNS theory claims that the three categories mentioned above are mutually exclusive, in the sense that the creative agenda of a coherent game can only fit one of the classes, and in particular should not change from one class to another, except maybe very slowly over time in a long campaign.

The justification of this is: Observations of Ron Edwards and some other theorists.

In particular, in my example that is a combination of character drama and mechanical fight scenes, one or the other should be in subservient role – either the blood opera acts merely as a justification for the fight scenes, or the tactics in the fight scenes are a mere means of resolution or expressing the characters’ inner natures.

Even supposing this is true, it is not particularly enlightening – how does one benefit from knowing that one of the activities is in some sense subservient to the other?

The Categories cover all of coherent roleplaying

Another claim of the theory is that there are no other creative agendae besides those covered by the three categories. Justification: Others have not been detected (by the people who have worked on the theory).

Coherence within every category

The theory claims that all priorities within a category are mutually compatible, or can be easily made so.

For example, some people in gamist play might enjoy strategy and negotiation and solving problems by creativity and strange use of spells. Others might enjoy character optimization and skirmish combat. While these goals might be compatible, it is also easy to imagine play where they are not. If game master prefers one and everyone else the other, than a conflict of some kind is very easy to imagine.

Similarly, right to dream is supposed to include both genre emulation (possibly via railroading) and heavy immersion in character. But railroading is often obvious, especially in the long run, and can thus destroy immersion when things get too contrived to be plausible. Coming up with other such examples is not difficult.

Non-emptiness of the agendae; or Simulationism

Right to dream has traditionally been poorly defined, and its existence as a creative agenda questioned, and thus uncertain. I admit to not knowing the state of the art here; my definition, above, is my own, and is probably flawed in one way or another.

More abstractly, this is a question of whether all the categories are non-empty.

GNS, in total

The problems of GNS come from claiming that it partititions the set of creative agendae neatly into three equivalence classes. Though this would be elegant, it is almost certainly wrong and in many ways. It also would not create significant, if any, practical value. If something is both useless and wrong, it should certainly be discarded.

The concept of creative agenda, and the role of GNS categories as families of examples, are still useful.

Classifying people and games

Creative agenda is a pattern of behaviour among people playing a game. Individual people might have preferences regarding the types of creativity they enjoy. Some people are more flexible about this, some less. It is practically always more useful to describe the specific creative agendae someone prefers, enjoys, can play with, or hates, rather than saying someone is a “gamist” or whatever.

Some games (as in published texts or products) are agnostic about purposes of play. This includes most traditional roleplaying games – they can be played with a large spectrum of different creative priorities. Some games are actively hostile towards some creative priorities, via a variety of means. (You can’t optimize a character without sufficiently complex character creation with several choices.) Some games make some patterns of creativity very easy to engage in. Some games have instructions for using them according to particular agenda; this might or might not be related to what the game is any good at. In short, it is quite useless to say that a given game is gamist. It is much better to say what specific thing it is good at, or makes difficult, or what it suggests in the text.

On creative agenda and GNS

Combat maneuvres in OSR

Tactics and strategy take place at three levels: The exploration level, the squad combat level, and the level of actions in combat.

  1. OSR is pretty unique among roleplaying games in being concerned about the exploration level.
  2. Many roleplaying games deal with the squad combat level, though only in the simple way of everyone deciding who they are going to attack.
  3. At the level of actions of individual combatant there are rulesets like Pathfinder, where you can optimize your character to be good at (say) tripping and dealing damage, and then in combat situations you can select which of them is more useful, or claim the game is poorly designed if neither is. There are also Burning wheel and The riddle of steel, which have more interesting combat minigames.

Here is a set of principles for adding level three to OSR play, without complicated lists of maneuvres. This same approach should work with any traditional roleplaying game which has characters acting as individuals in combat and their actions resolved separately. Tunnels and trolls requires a modified approach.

The purpose is not to reward players of describing cool stunts; but rather to reward player skill in coming up with good ideas and judging risk, which should naturally lead to more interesting combats. A hopeful side effect is an interest in (medieval) combat and the desire to find out more about it.

Core mechanic

Player states what their character is going to do. Someone, typically the referee, suggests how it should be mechanically resolved, so that the result is as realistic (within the setting and genre) as possible. Play group as a whole agrees to the suggestion or suggests a better one.

Over longer term play the solutions to common actions will probably become standardized, but this is not necessary.

This is precisely the same core mechanic as in all OSR play. Combat is in no way different from other parts of the game, at this level.

An example with a giant

You are playing an adventurer wearing gambeson, with a buckler hanging from your belt and a sheathed sword. Your group is surprised by a male giant, who runs out of the woods and smashes one of you allies, and is now looking for the next one to smash. What do you do?

Drawing the sword would take some time (say, -2 or -5 to next action, or some lost initiative if one uses individual initiative), but the buckler is pretty much instant to grab. Approaching the giant with either would be quite dangerous, as it has better reach and is prepared to attack; say -10 to attack roll if trying that. (An alternative would be some kind of attack of opportunity the giant can make; one can also let the player decide, here, how risky an approach they take.) On the other hand, waiting for the giant to (try to) smash someone else and running in after that would not carry such risks, but would give initiative to the giant and maybe cause another death. One might want to run between the giant’s legs and smash its crotch with the buckler; this could be a normal attack, and on success it causes hit point loss (not serious wounds, though) and forces the giant to save or be stunned for a round or two. Or, if the sword is drawn, stab the giant with that while running past, which would be a normal attack.

An example against a human

You are facing a swordsman. You have only a rondel dagger and a throwing knife. Both of you are unarmoured.

You might want to try to finish the combat with a single throwing knife – if it works, that is it, but if not, you are facing a swordsman while equipped with only a dagger, which is not a good situation. Maybe you pretend to throw the knife to see how the swordsman reacts. Or maybe you move close, and while almost as sword range, throw the knife (at a penalty or without a bonus) and charge in at the same time, trying to close the distance while your foe reacts to the knife. The enemy would have to choose between blocking/dodging the knife (while allowing you to close in), or hoping you fail at the throw and striking at you as you approach. On an excellent success (natural 20, margin of success 10+, etc.) they might get to do both.

If you miss with knife, then attacking a swordsman with only a dagger is similar to the situation with the giant, though hopefully less dire.


Whoever has initiative, decides how they will approach the situation. Others react, which might or might not give them choices. Above, the giant did not get to make any choices, while the swordsman did against the combined throw and charge.

I would suggest using individual initiative with this approach.

When is this appropriate

This type of resolution puts focus on the details of the combat. Some people might not want to do so, in which case they should not use this approach. Abstraction and focusing elsewhere is a valid choice, as always.

Combat maneuvres in OSR

Alignment and reactions

Assume alignment has a good-evil or a law-chaos axis and further assume that the rules system uses some sort of reaction rolls, with a good result meaning (more) favourable reaction and a bad result meaning hostile or negative reaction.

If using a law-chaos axis, then any lawful character has their reaction improved by one step if they assume the other party is lawful, and worsened by one step if they assume the other party to be chaotic.

If using a good-evil axis, then the reaction of good character to everyone increases by one step, while the reaction of evil characters to everyone worsens by one step.


A good society is a nice place. An evil one is not. But any good character might be taken advantage of and might not do as well as an evil character, especially in a situation of scarcity. This is a sort of prisoner’s dilemma. Law and chaos are mere tribes.

To determine the alignment of an established player character (that has been played for a while), consider:

  • How do they treat characters they assume to be lawful when compared to those they assume to be chaotic? If they clearly favour one side, then that’s their chaos-law alignment.
  • How do they treat complete strangers? If they mistreat them, they are evil. If they treat the strangers with caution, they are neutral. If they try to help and aid strangers, treating them as friends, they are good.
Alignment and reactions

Religious tenets for clerics

The typical presentation of a god in fantasy games assigns various spheres of influence to the god. In D&D 3 and Pathfinder the gods have associated domains.

Two tenets and one more

Each cleric, or other character drawing power from a more powerful being, writes two tenets. The referee must accept the tenets. Each selected tenet must be connected to a different domain associated with their patron god.

If the cleric chooses particular domains, then each chosen domain must have one associated tenet. Typical cleric selects two domains and so should write tenets related to those.

A tenet restricts the behaviour of the cleric, at least in some situations.

Each cleric also has the common tenet: to respect the god and what is sacred, to avoid what is profane, to participate in the daily rites and prayers, to be subject to the rituals associated with birth, death, coming of age, marriage, justice, etc. For monotheistic religions this tenet also includes not serving other gods.

Example: Tenets of Absolon

Though tenets are particular to a given character and god, some suggestions or defaults are useful.

  • Glory: Destroy the undead and the unholy, as well as mortals consorting with such.  Should this be beyond your present ability, harm them as much as you can and return with greater force.
  • Law: Expand the influence of Absolon, i.e. the laws and customs of the empire of Maldor.
  • Nobility: Treat nobles according to their station.
  • Aristocracy: Act and demand to be treated as one of noble blood. To select this tenet the character must be a member of a noble house in Maldor.
  • Leadership: Take leadership over your peers and inferiors, as defined by the hierarchy of the church of Absolon and empire of Maldor.
  • Sun: As the sun settles, call it back and keep its memory with an open flame; a campfire or more if you remain at a fixed location, a torch or greater flame if you are on the move.
  • Day: Act during the day, rest during the night.
  • Light: Never be in darkness.

Divine disfavour

Breaking a tenet accidentally, in a minor way (it is not noticed in play when it happens, only later), or against your will has the god choose some of your spells for you when you next prepare spells. The chosen spells should reflect the personal failing, rather than punish.

Acting against the god or breaking a tenet leads to the god withdrawing a measure of its power: the cleric loses their spells of highest level until they stop breaking the tenet. A full day of respectful obedience gives the powers back, as does using the atonement spell.

Denying the god, blasphemy, or severely breaking a tenet leads to immediate loss of divine powers. Repeatedly breaking a tenet, despite the warnings, leads to the same. Only an atonement may return the divine favour.

Changing tenets

Altering a tenet should follow the same guidelines as changing alignments, but should not be as big a deal. A ritual of some sort is also appropriate, or an atonement spell. The referee must accept the new form of the tenet.


The referee should always warn the player that their character is about to lose their divine powers, if such an event is about to happen. An act done without knowledge may only lead to the lesser consequences.

If a character is about to break a tenet or considers such an act, the referee should roll an easy check of wisdom or religious knowledge and, on success, warn the player of the consequences.

A player who asks about the divine consequences of an action should get an answer, possibly conditional on a check of wisdom or religious knowledge. The check should not, in general, be very difficult. Even low level divine spells (guidance, augury) should warn of divine disfavour with no chance of failure.

Design notes

The tenets are supposed to be somewhat restrictive, but not crippling, so they change gameplay without making the cleric unplayable. This is hard to tell without playtesting, but similar reasoning applies to selecting and defining BITs in Burning Wheel, keys in Solar system, etc.

The purpose of these rules is to emphasize the role of religion for adventurous religious characters and to make clerics of different faiths different in their behaviour. The ability to choose a number of tenets may mirror various factions among the faithful, and allows for greater freedom and more restrictive tenets that do not completely define or cripple a character.

The tie to domains or spheres of influence provides a starting point for defining the tenets, as well as a fixed number for their amount.

Religious tenets for clerics

Inheriting stats in generational D&D play

Suppose the particular stats (say, wisdom), of character’s parents are 12 (rolls 5, 2, 5) and 14 (rolls 5, 5, 4). How to determine the stat for the character?


This might be useful for generational play, or when starting a new game or campaign in a same setting but later in time. Playing as the children of previous, possibly retired, characters might add depth to the game.

Basic method

Stats are usually rolled with 3d6. We take first of those rolls from one parent, second from the other parent, and roll a new result for the third die. Using the example scores above: 5 from the first parent (rolled 1 with d3 to determine which die roll of the parent is used), 4 from the second parent (rolled 3 with d3), and new die score of 2; total is 11.

Special cases

If the stats of only one parent are known, then only one die is inherited, the other two rolled. If character is a result of asexual reproduction (with some variety, so not a perfect clone), then two separate dice are inherited from the single parent and one new die is added; this might be the dwarves of Dwimmermount or some insectile or vat-grown species.

If only the total score (such as 8) is known but the rolls are not, then any combination of die rolls with the correct sum is acceptable (for example 1, 3, 4). If one wants a little practice in calculating discrete probabilities, then one can figure out the probabilities of different combinations of scores by Bayes’ theorem and select one of them randomly.

Changes in stats due to adventuring

If a change alters the genotype of a character, including gametes, then the die rolls that constitute the stat should be changed accordingly. For example, if character is turned into supersoldier with strength of 20 and charisma of 2, then the corresponding rolls might be 6, 7, 7 and 0, 1, 1. It does not matter if they are not within the range of 3d6. Stat changes due to experience, age or practice should not influence the constituting rolls.

Exotic races

If the game uses stat requirements for non-humans, then any offspring not meeting the requirements dies before birth or shortly after it. A miscarriage or a stillborn child. If stat bonuses are used, then they don’t affect the constituting rolls of a stat and are applied after determining the base stat.


The new score of a child is affected by the scores of parents, but not determined by them. The expected value of a new score is the average of the following: first parent’s score, second parent’s score, and 21/2 (= 10 + 1/2 = 10.5). This gives regression towards the mean: no matter how bad or good the scores of the parents are, those of the child will tend to be somewhat average.

If the stats of parents were rolled with 3d6, and the stats of a child were rolled according to this method, then the scores of the child would be distributed as if they were also rolled with 3d6. This means that in the long run a population governed by this law (and no selection pressures) will have the stats of any given character being rolled with 3d6. This is true regardless of the stats of the first generation.

Inheriting stats in generational D&D play

More movement in combat

A completely untested house rule idea for Pathfinder, D&D 3, or other figure chess roleplaying games.

The idea

When defending in melee, the defender must make a choice before the attack: either take -2 AC or after the attack has been resolved move five feet away from the attacker, as though taking a 5′ step.

When attacking in melee, the attacker must make a choice before attack: either attack at -2 or after the attack has been resolved move 5′ towards the square the opponent occupied if there is space.

To select movement the defender must have space to move into. The attacker must have space to move if the defender moves or falls. Otherwise, -2.

The price

The rule would allow people to move farther than usual without attacks of opportunity by attacking or defending against incidental attacks. Hence there must be a limit.

I am not happy with the following solution, since it is inelegant. If the character about to move has movement left this turn (a move action, say), then the movement speed of the character is reduced. If they don’t have any (more) movement left, then the speed at which they could move next turn is reduced.

This should be considered and worded more carefully.


The rule only applies to melee attacks. Maybe avoiding ranged attacks and making reflex saves movement to random direction or -2 AC/reflex save?

The rule does not apply to attacks of opportunity.

For a medium-sized creature, moving away from an opponent means any of the three squares that are farthest from the enemy.

There will be unintended conflicts with various class abilities and presumably feats.


One may or may not consider this rule realistic. It would create more dynamic combat situations even with characters that have no special abilities. It would allow pressing through enemy lines or keeping a formation (at a price).  Having space behind you would be a boon.  Overpowering the enemy with sheer numbers might become somewhat easier.

More movement in combat