Prerequisite diagnosis

Find out that the thing you're stuck on isn't the thing you're stuck on; it's something from two years ago that never quite landed.

When to use this: When something feels wrong
Learning impact: Highest-leverage move
Learning evidence: Strong research tradition
Where you are with the topic: Building understanding
How often to use it: Worth repeating often

When to use it

You are stuck on this week's problem and suspect the real gap is earlier.
You re-read the chapter and still cannot say what is missing.

What to ask AI

Type something like: “I can't do …. Don't teach it to me. Instead:”.

Ask it to find gaps in what you wrote, without rewriting the whole explanation for you.

Keep the hard part on your side: you retrieve, attempt, explain, or decide before you ask for help.

How it helps

You cannot do integration by parts.

You do more integration problems.

None of it helps, and you conclude you're bad at calculus.

The actual problem is that you're shaky on the product rule.

Everything above it will fail, unpredictably and in ways that look like failures of the thing on top.

No amount of work on integration fixes it.

This is the commonest shape of being stuck in cumulative subjects, and it's almost impossible to self-diagnose, for a simple reason: you don't know which earlier thing the current thing depends on. That's exactly the knowledge you're missing.

How to run it

Name what you can't do, specifically. Not "calculus"; "I can't tell which part to call u.".
Ask for the chain, not the topic. What does this depend on, and what does.
Get diagnostic questions for each link, starting from the bottom. Short, one per link, designed to fail informatively.
Answer them honestly and in order. The temptation is to skip the easy ones. The easy ones are where the gap usually is: that's why it survived.
Stop at the first link you can't do cleanly. That's the real problem. Not the second one, not the topic you started with.
Next: fix that link, then climb. Re-test upward. Frequently three levels of difficulty dissolve at once, which is the characteristic experience of this method working.

Example exchange

You

I keep failing integration by parts. Don't teach me the method yet, diagnose what I'm missing underneath.

Start from the bottom. In one sentence, what is the product rule for derivatives?

You

Uh… multiply the first function by the derivative of the second? That doesn't sound complete.

That's the failed link, the product rule. Fix that before we touch integration by parts again.

Copyable prompt

I can't do [SPECIFIC THING]. Don't teach it to me. Instead:
1. list what it depends on, and what those depend on, three levels down;
2. give me ONE short diagnostic question for each, starting from the bottom;
3. wait for my answers;
4. tell me the lowest link I failed: that's what I actually need to fix.

The Tell

Here is how you know this method has flipped: you diagnosed a gap, fixed nothing, and went back to grinding the top-level topic. The diagnosis is worthless without the descent.

In that moment the AI (or the schedule, or the story) did the thinking, and you only recognised a finished product.

Tighten the prompt for prerequisite diagnosis so you attempt, decide, or retrieve before anything is handed to you.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated strong. Prerequisite gaps and knowledge dependence are central to cognitive apprenticeship and tutoring diagnosis: effective tutors locate the earliest broken link rather than reteaching the surface symptom (Wood, Bruner & Ross on contingent tutoring; Vygotsky’s the zone where you can almost succeed as the diagnostic target). Formative assessment traditions (Black & Wiliam) likewise treat “where you is” as the first move. AI probing that correctly isolates a prerequisite remains delivery-speculative, principle strength is high; automated diagnosis quality is not established by tutoring controlled trials in this corpus.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

The learning principle may be strong, but that does not prove an AI session delivers it well. Treat AI delivery as speculative unless a study of tutoring with this method is named, and none is claimed here.