Quest Ladder
TeachFORGE.ai
Quest Ladder / Mastery gate
The mastery gate: learned today, still known later
Two decisions — what proves they understand it now, and what proves they still remember it in a month.
Passing a question is not the same as knowing the topic.
What this tab is not
Mastery gate (optional for Quest Ladder) sets Skill status — whether this question type counts as current for a learner across every ladder they meet it in. It is not this visit's pass score, which lives on the ladder step.
The distinction matters. A learner can pass today's question and still not be marked as having learned the topic, because one correct answer with two hints is not evidence. That gap is exactly what this tab is for.
Two bars
Learned today
The first-time bar. Has the learner shown they understand this topic well enough to move on? It reads recent attempts: the average must clear the pass bar, no single attempt can be too weak, and at stricter levels the attempts must span more than one sitting without relying on hints.
Still known later
The memory check. When the schedule says a check is due, they must meet this bar again — usually on fewer attempts. It can be lower, the same, or higher than the first bar. Failing sends them back.
Levels, or your own numbers
The simple route is Follow my defaults, which uses the level you set in Quest Ladder settings for both bars. The alternative is Choose rigor for this question type, which asks you to pick strictness for each bar separately.
Five levels: Very loose, Loose, Standard, Strict and Very strict. A sentence under the chooser restates in words what the level you picked actually requires, which is worth reading — the names are relative and the sentence is not.
Most types should follow your defaults. Overriding per type is for the genuine exceptions: a safety-critical procedure that deserves Very strict, or a warm-up drill where Standard would nag a learner who has plainly got it.
The individual numbers
Behind the levels sit the numbers, and you can set them directly.
| Field | What it asks |
|---|---|
| Attempts that count | How many recent attempts the judgement reads. |
| Pass bar | The average those attempts must clear. |
| Lowest single attempt | A floor, so one good run cannot carry a bad one. |
| At least one attempt without hints | Evidence they can do it unaided. |
| Spread across different sittings | Evidence it survived a night's sleep. |
| Attempts for memory check · Memory-check pass bar · Lowest attempt (memory check) | The same three, for the later check. |
Two more shape how the attempts are combined. Score weighting chooses between Recent attempts weigh more and Smooth average (EMA). Gate calculation chooses Inherit from ladder/subject, Score only, or Score + confidence — the last of which will not pass a learner who is right but says they were guessing.
Only count fast timing bands restricts the evidence to answers given at speed, and Block over-confident wrong answers stops a confident mistake counting towards mastery.
When they are already strong
If they are already strong… decides what happens to a learner who plainly does not need the practice:
- Keep the normal path — nothing changes.
- Skip ahead when on a strong streak — they move on early.
- Offer one harder question first — they are stretched rather than released.
Strong streak length sets how long the streak must be. Both alternatives to the normal path are worth having: the first respects a learner's time, the second respects their appetite.
Try sample scores opens a small table of made-up attempts with their scores and hint usage, showing which combinations pass. Use it before you save. It is much faster than discovering the gate was too strict after a class has met it.