Updated September 2026 for Nousfarer.
A blank prompt asks a beginner to do several hard things at once: name the task, supply the context, judge the answer, and decide what comes next.
You may know what interests you without knowing how to ask for a useful lesson. Which ideas should come first? What background can it assume? What is a good question to try afterwards? Those are difficult choices when you are still finding your way into a subject.
What a chatbot asks you to decide
To get a decent lesson from a general-purpose chatbot, you have to specify most of it yourself:
| You decide in a chatbot | In Nousfarer, it comes from |
|---|---|
| The topic and its scope | The planet you land on and the institute you enter |
| Where to start | The institute’s introduction and its first room |
| What order to learn things in | The curriculum and its prerequisites, which set the room order |
| How to check understanding | The Sphinx’s three-question trial in each room |
| What to do after a mistake | A short pause and a way back to the reading |
You supply context by choosing where to go and what to open. You can start reading without writing instructions about lesson structure or question format.
But a choice identifies what you are studying. It says much less about what you already understand. Two people opening the same concept scroll can bring very different preparation. A topic and a place in a curriculum are a starting point; they are not the same as knowing the learner.
When the explanation assumes too much
Here is a small moment inside an institute. You read a concept scroll, take its trial, and pick a wrong answer. The answers lock for a few seconds (“Not quite. Try again in 4s”), and Review reading takes you back to the explanation while you wait.
On the second read you notice a term you skimmed past is doing important work. The sentence felt familiar, but you could not say why it supports one answer over another.
Rereading has found a gap. Now you can look the term up in an introductory textbook or a good reference, then come back to the same paragraph. With that preparation, the retry has a purpose.
This is the division of work. The game supplies a topic, an explanation, a question, and a way back to the reading. You still have to notice what you cannot explain, and decide whether trying again is enough.
Taking those decisions gives us obligations
When you don’t have to specify a lesson, the choices don’t disappear. We make them: the scope, the place in a sequence, the practice that follows. Those are educational decisions, even when they look like a room to enter or a button to press. So we hold ourselves to four standards:
- A framework introduction makes clear which approach you are entering.
- A concept article develops the topic it names.
- The route prepares you for the work ahead.
- The questions test what the explanation actually taught.
We don’t always meet them yet. An unexplained term, a question about a detail the reading never covered, or a question with more than one defensible answer is something specific for us to fix, and worth reporting. Structure also can’t prove an explanation is accurate; our trust post covers sources, human reading, and corrections.
There is still room for your own question
A prepared institute is useful when its scope matches what you want to learn. Sometimes you need a different starting point: a comparison, a narrower question, or a route that suits background the default order assumes you have.
At the ship’s Copilot station, describe that goal and what you already know. Copilot turns it into an expedition through what the Atlas covers. Read the proposed stops before setting off, and refine the request if the route misses your intent.
These two ways of starting belong together. Choose a place and study what’s there, or bring a question and let Copilot find a route. Either way, your effort goes into understanding, not into designing the lesson.
Enter Nousfarer, or read Articles & Quiz Trials for the practical details of studying inside an institute.