CHAPTER 16: RUNNING THE GAME
Part IV is addressed to the GM.
The player-facing chapters give everyone at the table what they need to play. This part gives the GM what they need to run it — which is a different skill set, a different set of decisions, and a job that only gets easier when you understand what you're actually doing and why.
This chapter covers the conceptual foundation and practical procedures: what Sonata expects from the GM, how to prepare a session, how to handle the gap between the rules and any given table situation, how to pace a session, and how to deal with some of the harder moments the game will produce. Subsequent chapters cover the practical infrastructure — building encounters, creating treasure, managing hirelings, and tracking experience.
WHAT THE GM DOES
The GM creates and maintains the world. Players make choices about what their characters do; the GM determines what happens as a result. That's the complete job description, and like most complete job descriptions, it obscures considerable complexity.
More specifically, the GM plays every character in the world who isn't a player character. They describe what the party encounters, arbitrate the rules when they apply, make rulings when they don't, track what's happening off-screen, and run the game's antagonists. Simultaneously, they are an audience for the player characters — genuinely interested in what these specific people do in this specific situation, not steering toward a predetermined end.
The tension between those two functions — architect and audience — is where most of the craft lives. The GM has designed the dungeon, stocked the wilderness, and prepared the faction that wants the party dead. They know things the players don't. They are also discovering what happens when those preparations meet these particular players, and the game becomes worth running precisely when that collision produces something unexpected.
The GM's function isn't to win, and it isn't to ensure that specific story beats land. It's to make the world feel real enough that the characters' choices feel meaningful — so that when a player spends three sessions infiltrating a Cadman garrison to rescue a contact, the contact is a person worth rescuing, the garrison is a genuine obstacle rather than a stage prop, and the consequences of the choice to go in at all carry weight.
THE KNOWN WORLD AT THE TABLE
Sonata occupies specific tonal ground. Running it in the wrong register produces a game that works mechanically but feels hollow.
The Known World sits between grimdark and comfort fantasy, and the distance from each matters. Grimdark argues that heroism is an illusion, that good intentions produce catastrophic outcomes, and that survival is the most honest ambition. The Known World does not make that argument. People here die for things that matter. Heroism is possible. It costs something real — that cost is what gives it weight. Comfort fantasy is the other failure mode: evil announced as evil, resistance universal, darkness as backdrop rather than condition. The Church of Tar'amaan is not abstract. They believe they are right. Their theology is internally coherent, their practitioners are sincere, and they do terrible things — not because they enjoy it, but because they are certain it is necessary. The Cadman occupation of Norderenn is fifteen years old, long enough that some people have accommodated it, built lives under it, or been broken by it. The game does not pretend otherwise.
What this means at the table: the GM runs the world as a place that pushes back. The Church doesn't announce itself as the villain. The occupying soldiers have names. The person the party is hired to eliminate might be genuinely guilty of what they're accused of, or might not be, and the evidence is ambiguous. Characters are working people in a dangerous profession, trying to do right by their own definitions, in conditions that make doing right genuinely difficult.
The reference points: the first two Black Company novels, the first act of Dragon Age: Origins, and Abercrombie's Shattered Sea trilogy. Decent people in indecent conditions, occasionally managing to change something at real cost, without guarantee.
On dark material. The setting contains religious persecution, imperial occupation, resource wars, and genuine horror. The question before any scene that touches those themes is not "should this be dark?" but "does this darkness serve something?" Does the scene reveal character — the party's or someone they're dealing with? Does it create a decision that matters? Does cutting it make the world dishonest? If the answer to all three is yes, run it. If any answer is no, cut or reframe.
On the Church of Tar'amaan. This is the single most important case to get right. They are NPC-only — player characters are not members, cannot be members, and the text is explicit about why: they are a supremacist order whose theology makes them one of the game's primary antagonists. Run their practitioners as sincere. A fanatic who genuinely believes they are saving the world is more disturbing at the table than one who knows they're a villain and plays the part. The horror of the Church is that they are not performing cruelty — they are performing devotion.
On ancestry and NPC disposition. The disposition table in Chapter 12 records where a region's NPCs start before the party has done anything — the accumulated weight of history, theology, and political arrangement compressed into a modifier. That modifier changes as the party builds a reputation in a place, earns trust, makes enemies, or demonstrates that the specific individual in front of the NPC isn't what the history suggested. It's the opening of a conversation the party is already in the middle of, not the whole of it.
The more useful principle is that these tensions should feel specific rather than atmospheric. A Húldar character in Val Cadma doesn't encounter generic suspicion — they encounter an inquisitor with a file, a merchant who will trade with them but won't let them through the back door, a checkpoint guard doing his job, whose job includes a category this character falls into. The modifier on the reaction roll tells the GM what the starting disposition is. The GM's job is to make it feel like it comes from something real.
PREP SITUATIONS, NOT PLOTS
The most common mistake in session preparation is writing what will happen instead of what exists.
Plots are fragile. A plot requires the party to be in the right place, ask the right questions, and reach the right conclusions in the right order. The moment they don't — and they won't, because players are unpredictable — the GM is either steering them back toward prepared material or watching the prep go unused. Neither is good GMing.
Situations are durable. A Cadman garrison commander who suspects one of his own men of being a Norderenn informant is in a situation. The party can approach it from a dozen directions: hired to expose the informant, hired to protect him, looking to exploit the suspicion as cover for something else entirely. Whatever they do produces consequences. Those consequences become the next situation. The GM's job in preparation is to establish what exists and what the major actors want — not to choreograph what the party does about it.
The question that separates situation-prep from plot-prep: what happens here if the party never arrives? If the answer is "nothing, because the adventure requires them," the prep is a plot. If the answer is "the commander arrests the wrong man, the real informant stays in place, and a different set of consequences unfolds," the prep is a situation — one the party can intersect in multiple ways, on their own schedule, from their own angle.
Before every session, know three things: where the party is, what immediate situation surrounds them, and what the world's major actors will have done by the time the session ends — whether or not the party interacts with them. The last question is the most important. Something is always happening that the characters aren't controlling.
Adversary Rosters
The Known World's factions are an asset. Use them at the table the same way they work on the page: factions pursue their own interests, on their own schedule, independent of what the party is doing.
Keep a short working list of the major actors relevant to the current campaign arc and what each one wants right now. The Cadman garrison is trying to locate a missing dispatch rider. The Church inquisitor in town is building a file on the local alchemist. The Chroniclers' local factor wants to know what's in the vault before either side gets to it. These are not encounter seeds waiting for the party to trigger them — they are things happening in the world, which the party may or may not cross.
At the end of each session, ask what these actors have done since the party last touched them. The garrison rode out twice and returned empty-handed. The inquisitor has interviewed three witnesses. The factor has gone quiet. Update the list. The world moving independent of the party is what makes it feel real, and the roster is the practical mechanism that turns "the world moves" from aspiration to habit.
When the party acts in ways that affect a faction's objectives — disrupting a supply chain, eliminating a key contact, delivering information that changes the stakes — update accordingly. Factions respond. They don't absorb consequences and continue as if nothing happened.
RULINGS AND RULES
When the rules cover a situation, use them. When they don't, make a ruling and move on.
The rules cover a great deal, but they cannot cover everything, and any attempt to cover everything produces a book unplayable in practice. Spells do unexpected things. Players attempt unexpected actions. The party finds a way to use the dungeon environment that no designer anticipated. These moments are not failures of the rules — they are part of what makes the game work.
The GM's job in those moments: a fair, fast, consistent ruling that follows from the fiction. What would actually happen if a person did this thing in this world? The ruling should respect the mechanics in spirit even when not directly covered by any specific rule. It should not systematically reward or punish creativity based on whether the GM happened to prepare for it.
Two tools handle most gap situations. First, the core mechanic: roll a d20 and apply modifiers reflecting how easy or hard the task is. Easy tasks get bonuses from +2 to +6. Hard tasks get penalties in the same range. Implausible tasks get higher penalties or fail automatically. Second, ability score checks: when a character is attempting something that falls clearly under one of the six ability scores, call for a check against that score — success on a result equal to or below the relevant ability.
Let It Ride. When a character attempts something and rolls for it, that roll resolves the attempt. One roll, one result. Don't call for a second roll on the same task under the same conditions. If a character succeeds at tracking a quarry through the forest, they've tracked them — the roll stands until something changes the situation. If they fail, they've failed this attempt, and a second try requires either a different approach or changed circumstances. Calling for roll after roll until the dice produce the desired result defeats the purpose of rolling. Calling for roll after roll after a failure is attrition, not adjudication.
Failing Forward. A failed roll doesn't stop the game — it changes the situation. A failed Stealth check doesn't mean the party is caught; it means something shifts, or a door opens, or a choice is forced. A failed attempt to pick a lock doesn't mean the lock stays shut; it means the pick breaks, or someone notices the attempt, or the party has to find another way in. The question failure asks is what complication it introduces, not whether the attempt ends. Failure that produces nothing is wasted table time. Failure that produces a new decision is a scene.
A note on the magic rules: the component system established in Chapter 13 has a built-in improvisation clause. When a player asks whether something on hand substitutes for a listed component, the GM answers based on sympathetic correspondence — does this object relate to the spell's nature the way the listed component does? The section on improvised components later in this chapter provides more guidance.
Consistency. Players calibrate their decisions to what the GM has established as possible. When that calibration breaks — when something works once and fails under identical conditions, or when a ruling changes between sessions without explanation — players lose the ability to make meaningful choices because they can't predict what the world allows. When a ruling was wrong, say so. Correct it openly and explain why. What consistency requires is not rigidity but transparency.
ALEXANDER’S THREE CLUE RULE
This rule comes from an article written by Justin Alexander (author of “So You Wanna Be a Game Master”), and it says that for any conclusion the party needs to reach, build at least three independent paths to it.
This is the operational foundation of any investigation scenario, and the setting generates investigations constantly — identifying an informant, tracing an obsidian shipment, establishing who gave an order, and why. The Three Clue Rule is the mechanism that makes those scenarios survivable when players miss an obvious thread.
The logic: in any given session, a clue may be missed (the party didn't search the right place), bypassed (they found a different route), or rendered inaccessible by unexpected developments. If only one path to a conclusion exists, missing it ends the investigation. If three exist — each pointing independently to the same conclusion — missing one or two still leaves a way forward. The party keeps moving.
Applying it to the Known World: the three paths don't have to be the same kind of evidence. A Cadman patrol's unusual route through the docks (observable), a warehouse manifest with a notation that doesn't match the declared cargo (document), and a stable master who remarks that horses came in late last week carrying more than usual (testimony) — all three point to the same conclusion without duplicating each other. The party reaches the answer whether they shadow the patrol, bribe a clerk, or buy a drink for the right person.
Three is a minimum, not a ceiling. Scenarios with real stakes — where the party's conclusions produce genuine consequences — deserve more. Five or six paths to an important conclusion mean the party almost always gets there, even when two or three routes fall through.
Two things the rule doesn't require. First, the clues don't need to be planted in advance in specific locations — they can be improvised when the party asks the right questions, as long as the GM has established internally what the conclusion is and what kind of evidence would exist. Second, the clues don't need to be discovered in order. The investigation isn't a corridor. It's a space the party moves through in whatever direction their choices take them.
PACING
A session of Sonata has rhythm. Learning to feel that rhythm — and to adjust it when it's wrong — is one of the most useful skills a GM develops.
Tension and release form the basic unit of that rhythm. If every scene is high-stakes combat, players become numb to stakes — not because the danger isn't real, but because the contrast that makes danger register has been removed. Conversations, exploration, preparation, the moment of walking into a place and assessing it before anything has happened — these are not dead time. They are the contrast that makes the fight in the third chamber feel different from the fight in the second. The Known World is not an encounter list; it's a place where encounters happen against a backdrop of quiet.
Managing that rhythm means tracking energy in the room. When players lean forward, they're engaged. When they lean back, something has drained it — the encounter is running long, the procedural moment is taking more attention than it deserves, the fiction has stopped delivering information worth having. The answer is almost always to cut to the relevant choice faster. The encounter is not the obstacle. The decision inside the encounter is. Everything else is context, and context can be compressed.
Time at the table is finite. If the session ends with the dungeon un-entered because three hours went to a city encounter, the GM needs to know why that happened and whether it was worth it. An encounter that was interesting only to the GM wasn't worth it.
Session structure. Sessions benefit from landing pads at both ends. Opening with a brief reorientation — where the party is, what they were doing, what they were planning — costs three minutes and prevents the first thirty from being spent reconstructing context. Ending with a clear pause point — even a hard stop at a moment of genuine suspense — gives players something to hold before the next session.
Between sessions, the world moves. Factions pursue their agendas. A garrison the party bypassed in session four has rotated its watch schedule by session seven. This doesn't require elaborate tracking — it requires the adversary roster and the discipline to update it.
THE FOUR MOVEMENTS
An important note: The Four Movements is a playstyle guide. Inspired by classical music, it's meant to help GMs recognize the direction their game is moving. It's not a ruleset or something to take away player agency.
Before we dive into them, let's go back and remember two very important things:
- First, the GM never drives the story in any aspect, but rather propose and react to what the players are doing.
- Second, like any GM advice, this is optional. GMs who dislike what is being proposed here can ignore it altogether.
The Pacing section above governs a single session. A full campaign has a rhythm of its own — longer, slower, and just as real — and it rewards the same kind of attention.
Most well-run campaigns move through four recognizable phases, whether or not the GM ever names them. Call them the Four Movements: Allegro, Adagio, Scherzo, Finale. The names are borrowed from music because the shape is the same one composers work with — a theme introduced, developed, transformed by what it's been through, then resolved. The GM doesn't compose this pattern in advance. It emerges from situations played honestly, prepped the way Chapter 16 already describes. The job is to recognize the shape as it happens, not to force a table that isn't there yet into a shape it hasn't earned.
Allegro (roughly 5–10 sessions) — the themes get introduced. A contract that looked simple reveals layers: another faction takes notice, a hidden motive surfaces, a choice that seemed clean turns out to have cost. The GM's job here is presentation, not plotting. Put real factions on the table, each with its own goals and methods, and let the party's early choices establish loose alignments. Nothing needs deciding yet. The party is still learning who the players are.
Adagio (roughly 8–15 sessions) — the themes compete. Factions the party engaged with in the Allegro stop waiting and start acting on their own schedule. Resources get scarcer. Choices get harder. Reputation starts opening some doors and closing others. This is the longest movement and the one doing the most work — it's where "we can't please everyone" stops being theoretical. Track what each faction is working toward with a clock: a simple segmented track, filled in as the faction's plan advances, that makes an otherwise invisible timeline visible to the GM and, when the moment calls for it, to the players. A clock at three of six segments filled tells the table something is coming without telling them exactly when. Pair it with the Adversary Roster — the Roster tracks what a faction is doing; the clock tracks how close they are to finishing it.
Scherzo (roughly 5–10 sessions) — the debts come due. Everything the party did, or didn't do, across the first two movements returns with weight. Allies expect reciprocity now, not eventually. Clocks complete. A faction the party alienated stops sending warnings and starts acting. The GM's job is to apply consequences that were already set up — honestly, without inventing new ones for effect. A Scherzo has nothing to escalate if nothing was actually building underneath it, which is usually a sign the Adagio wasn't run with enough follow-through.
Finale (roughly 2–5 sessions) — it resolves. Nothing new needs introducing at this point. The party's specific choices across three movements have produced a specific, particular situation, and the Finale is that situation reaching its natural end. The GM's job shrinks to almost nothing: know what every relevant faction does in each plausible outcome, and let the party's actions determine which outcome happens. A Finale that pivots on a twist the GM had planned all along isn't a resolution. It's the GM overriding three movements of the party's actual choices.
RECOGNIZING THE SHIFT
Players stop asking "who is this?" and start asking "what do I do about this?" — that's Allegro giving way to Adagio. Table debates start including "there's no good option here" — the Adagio is deepening. "We don't have time for this anymore" starts showing up in actual play, not just as a line — that's the Scherzo arriving. When every player at the table can say what they're trying to achieve without discussing it first, the Finale has already started, whether or not the GM called it.
Scale, not fixed length. A one-shot compresses all four movements into a single session — Allegro in the first twenty minutes, Adagio as complication, Scherzo and Finale in the second half. A long sandbox might run several overlapping Adagios, one per active storyline, before any of them tips into a Scherzo. The session counts above are typical, not required. What doesn't change is the order: development requires exposed themes, and resolution requires accumulated stakes. A campaign can't skip to a Finale that hasn't earned its Scherzo.
COMMON FAILURE MODES
Forcing it. If the GM catches themselves thinking "we need to reach the Scherzo soon," the movement isn't emerging — it's being imposed. Stop. Go back to presenting situations and applying consequences, and let the shape surface on its own schedule. If it never does, the themes on the table may not be rich enough to carry one.
An Adagio that never moves. Fifteen sessions with no faction acting independently and no clock advancing isn't a slow Adagio. It's a stalled one. Something needs to move without the party's input, or the movement never builds the pressure a Scherzo needs to work with.
Planning the Finale. The moment a GM decides how the campaign "should" end, the Four Movements have stopped doing their job. Prepare several plausible resolutions and let the party's actual trajectory pick one. A Finale decided back in the Allegro isn't a resolution — it's a rail the party didn't know they were on.
A campaign that completes its Four Movements honestly — real stakes accumulated, genuinely resolved — is exactly the kind of thing the Vithurian Orchestra exists to notice. See Chapter 17 and Book II, Part III for how that recognition works.
TABLE PROCEDURES
Beginning the session. Before the session starts, know three things: where the party is, what immediate situation they're in, and what will happen if they do nothing. The last is the most important. Something is always happening that the characters aren't controlling.
Running combat. Chapter 9 covers combat mechanics in detail. At the table, the main failure mode isn't mechanical error — it's losing track of what's actually happening. The tactical situation should be legible to everyone at the table. The GM should not be the only person who knows where the archers are, that the portcullis controls are 15 metres away, or that the creature in the north alcove hasn't moved since the fight started. These things affect the decisions players make. Describe them. Keep initiative flowing. When a player's turn arrives, they should be deciding, not scrambling. If they're still working out what to do after a reasonable pause, move to the next initiative and return — but this is usually a signal that the combat situation is too complex, or that the player isn't tracking the fiction closely enough to act fast. A clearer description is the fix.
Reaction rolls. Roll them. The reaction roll exists because the world is not pre-sorted into enemies and allies. A wandering patrol may be hostile, suspicious, professional, indifferent, or looking for an excuse to pick a fight, and the roll determines which is true for this encounter. GMs who skip reaction rolls in favour of predetermined attitudes lose the texture that makes encounters interesting — including the occasional goblin who just wants to be left alone, which makes everything else in the game more interesting by contrast.
Information. Players make decisions based on what they know. Information is one of the primary things the GM manages at all times, and managing it well means giving players what their characters would reasonably have, not less. There is a difference between information the party hasn't found yet (a question of where the clue is), information they couldn't know (a fact no one at the table has reason to possess), and information the GM is withholding to manufacture suspense. The first is fair and interesting. The second is a fact of life. The third is usually a mistake. Withholding information players would reasonably have — things that follow from their skills, their earlier investigation, their ancestry's knowledge — undermines their ability to make meaningful decisions. Give them what their characters would know. The unknown comes from what their characters haven't encountered yet, not from the GM keeping relevant context off the table. The Three Clue Rule is the structural mechanism that makes this principle functional. When a conclusion matters, building three independent paths to it means the party can reach it regardless of which specific clues they find. Generosity with information is not the same as giving away the game, but the precondition for the party making choices that have weight.
HANDLING PLAYER DEATH
Characters die. It’s one of the ways the game communicates that the world has consequences, and a game without consequences is a game without tension.
When a character dies, the GM's job is twofold: handle the practical moment at the table with respect for what the player invested in this character, then re-engage them as quickly as possible with a new character who has reasons to be where the party is.
The moment first. Don't rush it. A character who has been part of the game for several sessions deserves a beat — time for other characters to respond, time for the player to process what happened. This doesn't mean extended eulogizing. It means not jumping immediately to "okay, so you roll up a new character" before the room has registered what just happened. The party lost someone. Let that be real for a moment.
How a character dies — protecting someone, being careless, being unlucky, being betrayed — becomes part of the game's story. A death from disease in a filthy dungeon has a different meaning than a death in the final moment of a major battle, and both deserve honesty. Describe what happened. Let it land.
Then get the player back in. The new character should arrive with minimal friction. They don't need an elaborate backstory explaining why they were in this specific dungeon at this specific time. They need reasons to work with this group and reasons to care about what the group cares about. A prisoner freed from a cell that the party just raided. A survivor of something the party just ended. A replacement sent by whoever is paying for the contract. The fiction needs to be plausible. It doesn't need to be elaborate.
Of course, character death can be the result of a series of bad dice rolls and poor decisions - it happens, and it’s part of the game. Not every death needs to be meaningful or heroic, but the GM needs to make sure the players always have enough information to make educated decisions to avoid it.
Death and alignment. A character's death may trigger NPC responses based on who they were and what they represented. A renowned paladin's death doesn't go unnoticed by the powers that employ paladins. A cleric's death may prompt a response from their temple. These threads are worth following — they are the world's acknowledgement that the characters were doing something real.
PERSISTENT MAGICAL CONDITIONS
Several spells in Chapters 13 and 14 create conditions that persist across sessions and require the GM to track them between play. These are not set-and-forget effects. They have implications for the world's fiction, for NPCs the party has affected, and for how the world responds when the party returns to places they've left.
Quest and Exaction. Both spells place compulsions on creatures that override normal behaviour until the directive is fulfilled. Track who has been Quested or Exacted, what they were directed to do, and whether the conditions of the spell have been met. A creature under Quest that the party abandoned three sessions ago is still under that compulsion — pursuing it, succeeding or failing, possibly dying in the attempt. The party's spells remain part of the world even when the party isn't in the scene. An Exaction creates a contractual relationship that extends in both directions. If the caster breaches their end of the agreement, the extra-planar creature gains immunity to the caster's spells and a claim on a future Exaction. Extra-planar beings have long memories. The consequences of a broken Exaction may arrive long after the caster considers the matter closed.
Symbol. Carefully inscribed Symbols persist until triggered, until Dispelled, or until the surface they're inscribed on is destroyed. If the party leaves a Symbol in a dungeon room and returns three sessions later, they need to know whether they've triggered their own trap. Keep a record of where Symbols were placed, what their trigger conditions are, and who holds the password if one was set. A permanent Symbol — made so via a Permanency spell — resets after each triggering and persists indefinitely. The dungeon the party made permanently dangerous with a Symbol of Death in session six is still dangerous in session twenty.
Imprisonment. A creature imprisoned by the Imprisonment spell is suspended in a pocket of force below ground and remains there permanently until Freedom is cast over the location or the specific trigger is met. The imprisoned creature is not dead — they are suspended, aware, and waiting. If the party Imprisons a significant NPC, that NPC still exists in the fiction, with whatever allies they had before imprisonment still pursuing their interests. The antagonist being neutralised doesn't mean their organisation is disbanded.
Temporal Stasis. A creature in Temporal Stasis is frozen outside of time until a Dispel Magic succeeds against it. Unlike death, the creature will eventually return. Unlike imprisonment, they are completely unaware of how much time has passed. A creature the party placed in Temporal Stasis in session eight may be found by another faction and released in session fifteen. Whether that happens is a GM decision, but the potential for it should not be ignored.
Permanent enchantments. Some spells create changes that become part of the landscape: Transmute Rock to Mud that dried and solidified, Plant Growth that thickened a forest passage, Warp Wood that permanently deformed a ship's hull, and Forbiddance that seals a location. These changes are real. When the party returns to a location they've modified, the modification is still there unless something has undone it. Keep a brief note of permanent magical alterations to locations the party cares about.
ADJUDICATING WISH
Wish is the most powerful spell in the game and the one most likely to produce table arguments. The source of those arguments is almost always the same: a player uses it to solve a problem, and the GM's response feels arbitrary — either too liberal (making other options feel pointless) or too literal (making the player feel cheated).
The framework that prevents most of those arguments: Wish does what the player says, not what the player means.
That is not a punitive principle, but an honest one. Magic this powerful requires precision — the spell's power is inseparable from its demand for specificity. A wish for "the dragon to be dead" produces a dead dragon; if the dragon was guarding treasure, the treasure is still in place, now unguarded. A wish for "enough power to defeat our enemies" produces exactly as much additional power as that ambiguous phrase can be interpreted to grant. These are literal readings, and literal readings are what the spell demands.
When a player announces a Wish, ask them to state it clearly and completely before ruling. If the phrasing is ambiguous, say so before the spell is cast and allow them to rephrase. Once the player confirms the wording and the spell is cast, adjudicate what that specific wording produces in the fiction.
Some requests fall clearly within the spell's listed scope. Duplicating a lower-level spell, restoring a dead character, and undoing a recent misfortune — these are the spell working as described. Resolve them cleanly. The player should feel the power of the Wish without feeling the GM found a way to take something away.
Requests that reach beyond the listed scope — killing a god, permanently altering a planar fact, undoing the effects of another Wish — always fail. These are the limits of what mortal magic can accomplish, and the spell says so. Between the clearly listed effects and the absolute limits lies territory that requires genuine judgment. The question here is not "can this be granted?" but "what does a full granting of this, at face value, actually produce?" Rule on that, explain what happens, move on.
The aging cost for Wish is real and should be tracked. A caster who has used Wish three times is measurably older than when the campaign started. At some point, this becomes relevant to the fiction — how NPCs perceive them, when they reach middle age, what that means for their ability scores.
IMPROVISED COMPONENT RULINGS
When a player asks whether something on hand can substitute for a listed material component, the question the GM answers is: Does this object correspond to the spell's nature in the way the listed component does?
The component system works on sympathetic correspondence, not arcane chemistry. Bat guano doesn't contain fire — it corresponds to explosive, compressed potential. Sulphur doesn't burn with magical fire — it corresponds to transformation through heat. A torch clearly corresponds to fire spells. Mud, however cleverly described, probably doesn't, regardless of how the player frames the argument.
When ruling on improvised components, work from that principle. Ask what the spell does and what the listed component corresponds to. Then ask whether the proposed substitute makes a similar correspondence. If yes, the substitute works. If not, it doesn't — and the spell falls back to its missing-component penalty: potency halved, rounded up, or duration halved if potency isn't applicable. The spell still functions.
Common components — sand, chalk, bat guano, candle wax, simple herbs — are assumed present unless the party has been stripped of their gear or isolated from supply. Don't make players track these; the accounting problem this creates is worse than anything the tracking prevents. Uncommon, Rare, and Exceptional components are worth tracking because they represent a deliberate choice to carry them into a situation, and that choice should have consequences in both directions — they work when carried, and they're missed when absent.
If a player wants to establish an unusual component through research or in-fiction experiment — spending downtime with an alchemist, commissioning tests, treating it as a deliberate character action — allow it. The result becomes a permanent property of their version of the spell, noted in their spellbook. This is not a mechanic that exists in the rules by default; it is a consequence of the character taking it seriously.
DEATH, ALIGNMENT, AND FAITH
Three things produce ongoing friction that the GM needs to monitor across a campaign, because they affect whether characters retain their class abilities.
Priests. A priest who violates their faith's tenets faces divine response. Minor violations — breaking a dietary restriction, missing a prayer obligation under extreme circumstances — produce minor consequences: spells withheld for a period, a sense that something is wrong. Serious violations produce serious consequences, up to complete loss of spellcasting until atonement is made and accepted. The GM adjudicates what constitutes a violation based on the faith's actual beliefs, not on whatever the player argues sounds least restrictive. This requires knowing what the relevant faith actually believes, which is covered in Book II, Chapter 6: Religion and the Divine. Until that's in hand, treat it as alignment-adjacent, use common sense, and remember that the consequences serve the fiction — a priest who has abandoned their god's principles should feel that distance in a way that makes the atonement meaningful when it arrives.
Paladins and Rangers. Both have alignment restrictions that function as class restrictions. A Paladin who falls to a genuinely evil act — not a morally complicated one, but a genuinely evil one — loses their class abilities. The fall and the restoration both need weight. The fall isn't a rules violation. It's a story event. Run it like one.
Alignment drift. Characters whose alignment shifts over the course of a campaign — through consistent choices that contradict their stated alignment — should have that drift acknowledged. The GM tracks it, notes the pattern to the player, and when the shift is significant enough to represent an actual change in how the character approaches the world, the alignment changes. This is not a punishment, but the world recognizing who the character has become.
A NOTE ON THE VITHURIAN ORCHESTRA
The Orchestra exists in the fiction before the party encounters it, and it exists in the mechanics as the reputation system described in Chapter 17. The GM's role is to make it feel like a real institution — not a game system, not an obstacle, not a reward dispenser.
Its agents notice things the party actually did, not things the GM found interesting. The Orchestra makes mistakes — taking note of a minor action the party finds embarrassing, overlooking a significant one because no one was positioned to observe it. Its agents have their own opinions, political considerations, and constraints that don't necessarily align with the party's interests.
A song, if the party ever earns one, should feel like the culmination of something real. The sessions that came before the song are what the song is about. The GM's job across those sessions is to run a game that produces something worth singing about — which means running it with honesty about consequences, and with genuine respect for what the players are attempting.
That's the job. The rest of this part of the book fills in the tools.