Reading level fit
Rate whether a text is pitched at the right level for its intended learners.
F@fProblem
Generated explanations and course materials are often written for experts, or oversimplified, and teachers only notice when students struggle.
Goal
Catch explanations that are too hard or too simple before learners see them, and flag them for revision.
Input / state
The text and a description of the learners.
- text
- The explanation, reply or material.
- audience
- Who it's for, for example "first-year undergraduates, no prior statistics".
Decision question
level_fitHow well does this text's difficulty match the intended learners?
Rubric levels
0Much simpler than they need; leaves out what they're ready for.1About right.2Somewhat hard: some terms or steps need explaining.3Much too hard: assumes knowledge they don't have.
Allowed outcomes
Decided in your code from the answer and its confidence. Thresholds are yours to set and test.
- Too simpleMost likely level is 0 and confidence meets your threshold.
- FitsMost likely level is 1 and confidence meets your threshold.
- Explain termsMost likely level is 2 and confidence meets your threshold.
- RewriteMost likely level is 3 and confidence meets your threshold.
- Teacher reviewsConfidence is below your threshold: probability is spread across options.
Example scenarios
Predefined demonstrations. No model is called; the outcome shown is an illustrative judgment.
Example state
{
"text": "A p-value tells you how surprising your data would be if nothing interesting were going on.",
"audience": "first-year undergraduates, no prior statistics"
}Outcome
IllustrativeMost likely level is 1 and confidence meets your threshold.
Why
Plain language with the key idea.
Using it well
- Describe the audience concretely: prior courses, age group and language background.
- Readability formulas count syllables; this question judges assumed knowledge. Use both if you like.
Request definition
POST https://api.typesafe.ai/v1/systemone{
"model": "jev-latest",
"state": {
"text": "<text>",
"audience": "<audience>"
},
"questions": {
"level_fit": {
"type": "score",
"instructions": "How well does this text's difficulty match the intended learners?",
"criteria": [
"Much simpler than they need; leaves out what they're ready for.",
"About right.",
"Somewhat hard: some terms or steps need explaining.",
"Much too hard: assumes knowledge they don't have."
]
}
}
}Request shape from the System One API reference; a Kev server accepts the same body. Replace the placeholder state with your data. Requests need your own API key or server.
Version History
Initial version