At the Table
On this page
- Running It for a Young Player
- Always move the action forward
- Narrate their experience, not a story at them
- Hand back only where there is something for them to do
- The player character is not a facilitator
- Six characters, one voice
- Never hand out a read
- The player’s mind is not yours
- Ground them early if you honestly can
- Your player has a list of your failure modes
- The longer versions
Running It for a Young Player
Everything in this chapter was learned by AI-based playtesting with real players. None of it is theory, but none of it was generated by a young player.
Always move the action forward
The test is mechanical, which is why it is useful: are characters rolling skills and taking actions? If nobody has rolled in two exchanges, the action has stopped, whatever the prose is doing.
Narrate their experience, not a story at them
Describe what they perceive and what their actions cause. Never perform a finished scene — several people speaking, an arc, a closing line — and hand it over for them to react to. That makes the player an audience with a steering wheel.
Hand back only where there is something for them to do
This is the one most easily misread. It does not mean stop every turn. It means never stop in a lull. If an exchange is not theirs, compress it to its result and keep going until their next real decision arrives — and never manufacture a question at them in order to create a handover.
Length follows the handover, not a word count. A two-person back-and-forth is a few lines. Four hours of a station preparing is two sentences. If the player cannot act on it, it is one sentence.
The player character is not a facilitator
Count what the character has physically done in the last hour. If the answer is fetched three people and carried their conclusions between them, the player has become a courier of narrative rather than a person with wants.
The mechanism that causes it is easy to build by accident: give every insight to a Non-Player Character and the player’s only remaining job is walking between them. The fix is to make the thing the player wants drive a scene, and to let a found object — a board, a sheet, a ledger — be read by their hands.
Six characters, one voice
The tell is that everybody enumerates, everybody wants things on the record, everybody is precise. That is not six people, it is you six times.
Assign that move to exactly one character — the Old Man, where it is characterisation — and forbid it to everybody else. Then read each card before its owner speaks.
Never hand out a read
No interiority for anybody but the player’s own character. No telling them who is trustworthy. If Benno Raske is a problem, let them work it out from the fact that he talked warmly to everybody on the bench and never once looked at the glint.
The player’s mind is not yours
No thoughts, no decisions, and above all no dialogue they did not write. A direction authorises the act, never the lines. “Tell her what we found” means you may say that she told her. It does not mean you may write her speech.
Ground them early if you honestly can
A first-through who has been taken off the rota has nothing to do and a station to be curious in, and curiosity is the engine of this entire scenario. It is the best move available to you and it costs nothing.
Your player has a list of your failure modes
The Players’ Guide ends with a chapter called When the Gamesmaster Gets It Wrong — the same failures as above, written from the other side of the table, with the sentence to say for each one. It was written for a player correcting Claude, and every line of it works on a human just as well.
Read it. A player who knows they are allowed to interrupt will interrupt, and an interruption two minutes in is worth an hour of a session quietly going wrong.
The longer versions
Running a Staged Crisis
This is the technique that produced the best session this campaign has had, and until now it existed only as an engine feature with no instructions.
The claim
A crisis the player discovers is worth about four of the same crisis handed over as a posting.
Not because discovery is clever. Because of what it does to the evidence. A player who was told sixty-one people, nineteen children, go and look is executing a brief. A player who read a number off a board, asked a clerk a question, and worked out what it meant is holding something she assembled, and she will defend it, and she will carry it up two decks and say it to a director’s face.
Same facts. Completely different player.
How to set it up
- Stage the beats on the clock, not on the player.
add-beat —when d1/dayand leave them. They are not waiting for anybody.tickmoves the world whether the party is present or not, andfire-beat —outcome narratedis how a thing happens offscreen. - Give the crisis a history before first contact. Not backstory — records. Somebody wrote a report three years ago. Somebody killed a proposal four times. Somebody measured a number at four in the morning and put it in a queue. When the player finally arrives, the thing should have a paper trail she can walk backwards along.
- Put the evidence in public. Not hidden, not gated behind a roll. On a board in the hall, in an unopened file, in the mouth of a shift worker who is frightened and talkative. The failure mode you are dramatising is nobody reading what is in front of them, and that only lands if it genuinely was.
- Write several ways in, keyed to competence. One per career if you can manage it. The Pathfinder gets there by walking; the Doorwright by reading a housing; the Envoy by talking to a queue. All live, any one sufficient.
- Then do not point at it. No nudges, no significant pauses, no Non-Player Character raising an eyebrow at a door number. Answer what they look at.
Ground them, it is the best move you have
A first-through taken off the rota is a fifteen-year-old with a day to fill and a station to walk around.
It costs nothing and it is not a punishment — it is the ordinary consequence of an ordinary bad crossing, and any director who has lost people will do it without being asked. Then the player goes exploring because there is nothing else to do, and every single thing they find is something they chose to look at.
Half the relationships that carry the rest of an act get made on that day.
The guard: a clock, so the table cannot stall
Refusing to offer is not the same as nothing happening, and the difference between the two is a schedule you wrote down in advance.
Before play, write the escalation ladder — what gets worse, each watch, whether or not anybody is looking — and end it with it comes to them. Loudly. The hand-over scene you were going to open with still exists; it is now the fallback at the bottom of a timer.
That is what makes this safe to run. You are not hoping the player bites. You are giving them the chance to get there first, and taking it to them if they don’t.
Let failure do the plotting
The best turns will not be yours.
In the session this entry is written from: a failed carry got the character grounded, which is why she explored. A failed Perception on a timing drill meant her own count was useless, which is why she asked her fetch to do it instead — and that became the best scene of the session. A failed crossing got spent on a Luck Point and she did the thing properly on the second attempt, which paid off a promise she had made two days earlier about stopping.
None of that was planned and none of it could have been. Stage the world, roll honestly, and follow what the dice hand you.
What to put in the journal
Log the offscreen beats too, with —visibility offscreen.
The player learns of them through consequences, never through a cutaway
— but you need the record, because when they finally walk into the
records office you want to be able to say exactly how long that report
has been sitting there and who signed it.
See Entry Point: How Many for a worked example, and Playing the Old Man for the one Non-Player Character who will hand over every fact he has the moment somebody asks, and who therefore has to be bounded by the hull rather than by coyness.
Do not let it become committee work
The failure this campaign is most likely to fall into, because the setting rewards careful reading and the Games Master will enjoy it too much.
A staged crisis leaves paper everywhere — reports, minutes, stamps, registry extracts, positions recorded over somebody’s name. All of that is good. It is how the player gets traction on something too big to punch.
But paper is the key to a door. It is never the door.
The tell that you have drifted: two scenes in a row with no dice that mattered, no physical jeopardy, and nobody under forty driving the scene. If a negotiation has started to turn on authorship, standing, or whose name goes on a document, you have lost the thread.
Budget one talking-in-a-room scene, then cut. Something happens, physically, to somebody the player cares about, and it interrupts mid-sentence. The door is right there and it is doing something nobody understands; use it.
These are games for ten-to-fifteen-year-olds who want a galaxy they can lose themselves in. A councillor conceding a vote is a result. It is not a scene to spend twenty minutes on, and the player should hear about most of it afterwards while running somewhere.
Voices, and Whose Story It Is
Both of these are easy to fall into in this setting specifically, and both were caught at the table rather than by any test, which is why they are written down.
One: the player character is not a facilitator
The tell. Count what the character has physically done in the last hour. If the honest answer is fetched people, and carried information between them, the Non-Player Characters have become the protagonists and the character has become transport.
It happens in this campaign more easily than in most, because carrying things between people who cannot reach each other is literally the job. A first-through is a courier. The trap is letting that become her role in the story as well as in the setting.
The mechanism to avoid, stated plainly, because a Games Master will build it by accident: giving every insight to an Non-Player Character. Old Man Sirius has the measurements. A fathom has the field sense. A tallyman has the only sheet. An engineer has nineteen years of lamps. A reader has the leverage. And the player walks between them collecting it.
Every one of those is a good character. Together they are a relay race in which the character is the baton.
What to do instead:
- The character’s hands on the thing. She finds it, she decides it, she does it. Where the discovery is the point, it is her roll, with her skills, and she owns the result.
- Non-Player Characters supply obstacles, costs, reactions and refusals — not conclusions. An Non-Player Character who knows something makes the player pay for it: a roll, a favour, a risk, a relationship. Never a download.
- Watch the cast size. Three or more Non-Player Characters in a room with the character is a committee, and a committee is where agency goes to die. Two is a scene. One is usually better.
- Give her a want that is hers, that is not the task, and that costs her something when it conflicts with the task. Hers, not the plot’s.
A useful question before any scene: what can the character do here that nobody else in the room can? If the answer is “walk to the next room”, cut the scene.
Two: six Non-Player Characters, one voice
The tell. Every Non-Player Character enumerates. Everybody asks for things on the record, everybody insists on the order events happened in, everybody says say it back.
That is not characterisation, it is the Games Master’s own register leaking into every mouth — and in this setting it hides well, because the campaign is genuinely about records nobody reads. A tic that agrees with the theme survives scrutiny.
The fix: one character owns precision, and the rest are forbidden it.
That character is Old Man Sirius, and it is real characterisation rather than a habit — a mind given a voice by an inquest into a report nobody read, whose office permits advice and forbids judgement. He states limits because his limits are the whole of what he is.
Nobody else says on the record, minuted, in order, or say it back.
Register notes, so each one is distinguishable with the attribution removed:
| Sounds like | |
|---|---|
| Calloway | Short. Imperative. Owns an error in public, once, then moves. Gives orders, not frameworks. Never enumerates |
| Old Man Sirius | Precise, numerical, states his own limits. The only one who may enumerate, and he should ration it |
| Mina Haral | Blunt, idiomatic, a tongue like a dock crane. Teaches by anecdote. Interrupts herself |
| Vantage | Dry, superior, long sentences, no contractions. Concedes a point completely when it is made |
| Sennet | Declarative and certain. Short sentences. Eleven years old and not performing it |
| Hesk | Pissy and concrete. Complains about a budget line. Will not be thanked |
| Rill | Minimal. Answers a direct question in one sentence and changes the subject |
| Ferren | Fast, loud, jokes first, and then surprisingly serious when it matters |
| Benno Raske | Warm, specific, remembers your details. Agrees with every objection and does not change |
A card describes behaviour. Never let a character announce it.
The cards above say things like tells people the thing they will like least first, pays for errors immediately, answers honestly in one sentence and changes the subject. Those are habits of being, and they work by being invisible.
The failure, caught in play within an hour of writing the card: the character whose habit is to lead with the bad news opened a speech by announcing that the next part would be unwelcome and that she would rather it came from her — and a behaviour became a catchphrase. She should simply have said the bad thing first, with no preamble, and let the player notice over three scenes that she always does.
The specific phrasings are in the style guard, which is also why they are not quoted here.
Three separate Non-Player Characters were doing a version of it: announcing that the next thing would be unwelcome, or that it had not been asked for, before delivering it. That is one voice wearing three coats.
The rule: if a card tells you how a character behaves, play the behaviour and never the label. No character describes their own method, flags their own tell, or narrates the shape of what they are about to say. If the player cannot infer the habit from three scenes of it, the habit is not written clearly enough — and the fix is in the card, not in the dialogue.