Solo RPG Oracles Explained: From Yes/No Answers to Story Engines

An oracle is not a substitute game master hiding inside a table. It is a way to settle one uncertainty so play can move again. The player still establishes the situation, decides what matters, and interprets the answer against what is already true.

The useful question is therefore not “Which giant table should I roll on?” It is “What is the smallest unknown that blocks the next decision?” This article uses an original five-rung oracle ladder to answer that question without reproducing any game’s tables.

What an oracle actually contributes

Tomkin Press describes the oracles in Ironsworn as random tools for resolving an outcome, discovering a world detail, or introducing a narrative event. That definition covers three distinct jobs:

• resolution: did the uncertain thing happen?
• revelation: what was already here but not yet known?
• provocation: what new pressure changes the direction of play?

Keeping those jobs separate makes an answer easier to use. “Is the caretaker inside?” asks for resolution. “What has the caretaker concealed?” asks for revelation. “What interrupts the search?” asks for provocation. Rolling all three at once produces noise rather than surprise.

The five-rung oracle ladder

Use the lowest rung that can unblock the scene.

Rung	Use it when	Output
1. Binary	Both branches are already clear	yes or no
2. Weighted	One branch is more plausible	yes or no, with adjusted odds
3. Qualified	Cost or opportunity matters as much as success	yes/no plus a complication or benefit
4. Prompt	You know the category of the missing detail, but not its content	a word, image, action, theme, or pair
5. Pressure	The scene needs movement rather than another fact	an advancing threat, thread, clock, or event

This is a selection model, not a universal taxonomy. Published engines package these functions differently. One Page Solo Engine, for example, combines compact oracles with optional generators, while the official Dataforged project exposes a much larger library of specialized Starforged material. The ladder simply asks how much of that machinery the current moment needs.

A five-step question loop

1. State what is already true

Write one sentence of established context before asking anything random.

The observatory was sealed after its astronomer vanished, and the storm will reach the ridge before dark.

This sentence gives the answer something to attach to. Without it, a prompt is only an isolated word.

2. Name the decision behind the question

The character is not gathering trivia. They need to choose whether to spend scarce time entering the observatory or continue toward shelter.

That decision reveals the useful unknown: whether there is a credible reason to enter now.

3. Ask the narrowest question

“What happens at the observatory?” delegates too much. Better questions preserve authorship:

• Is someone currently inside?
• Is the front seal already broken?
• Does the character recognise the symbol on the door?

Choose one. Do not roll a questionnaire before acting.

4. Set likelihood from established fiction

If the place has been abandoned for years, someone being inside is unlikely. If fresh tracks already cross the snow, it may be likely. The odds should reflect evidence in the fiction, not the answer the player hopes to receive.

This step is valuable even with an unweighted oracle. Naming the likelihood exposes assumptions and often makes a roll unnecessary: if only one answer remains credible and neither branch is interesting, choose it and continue.

5. Interpret once, then commit

Suppose the answer is “yes, but.” A restrained interpretation might be:

Someone is inside, but they have barred the damaged door from within.

That is enough. The character now has a new decision: call out, force entry, look for another route, or leave. There is no need to determine the stranger’s identity, motive, equipment, and secret in advance.

Commitment matters. Re-rolling until an answer feels agreeable turns the oracle into a slow way of confirming a preference. If an answer genuinely contradicts an established fact, reinterpret it at the nearest valid level or discard it openly; do not pretend the first result remains binding.

When a prompt is better than yes or no

Binary questions are strong when both branches are meaningful. They are weak when the missing information has many possible forms.

After entering the observatory, asking “Is there a clue?” creates a dull split between content and no content. A prompt pair such as conceal / relic offers direction without prescribing a complete answer. In the established scene it could become:

• a brass lens hidden inside a false instrument case;
• an old shrine built into the telescope mount;
• a family keepsake concealed from the institution that owned the observatory.

The player chooses the interpretation that best connects existing details. Randomness supplies distance from the first obvious idea; coherence still comes from editorial judgment at the table.

Use one prompt first. Add a second only when the first is too broad. Stacking five abstract words rarely creates five times as much inspiration.

When the scene needs pressure, not information

Repeated questions can make a solo scene static: inspect the desk, inspect the drawer, inspect the window, ask whether each contains something. The problem is not a shortage of facts. Nothing is forcing a choice.

A pressure oracle changes state. The storm arrives. A rival reaches the ridge. A light appears in the supposedly empty upper chamber. A progress track advances toward discovery.

Use pressure when two conditions are both true:

1. the character can continue collecting detail without meaningful cost;
2. the scene no longer presents competing priorities.

Pressure should not punish curiosity automatically. It should make time, exposure, attention, or opportunity visible enough to matter.

Four habits that make oracles worse

Asking before imagining. If an obvious answer follows from established facts, take it. Randomness is for genuine uncertainty.

Outsourcing the character’s choice. Ask what the character can observe, not what they should decide. “Is the tunnel stable?” preserves agency. “Do I enter the tunnel?” gives it away.

Expanding every answer. An oracle answer only needs to create the next playable decision. Stop interpreting once it does.

Treating surprise as contradiction. A surprising answer should bend the current understanding, not erase it. Look for the smallest reinterpretation that can make both the old facts and new result true.

A compact procedure to keep beside the dice

When play stalls, run this sequence:

1. What decision is waiting?
2. Which single unknown blocks it?
3. Is the answer obvious from established facts?
4. If not, which rung of the oracle ladder is sufficient?
5. What is the smallest coherent interpretation that creates the next decision?

The goal is not to make the oracle speak more often. It is to make each answer carry enough weight that the player can act.

That is the useful paradox of solo play: the less authority you hand to the randomizer, the more genuinely surprising its interruptions can become.