What Should You Do When You’re Stuck on a Problem?

When you are stuck on a problem, do not choose between staring at it forever and immediately reading the full answer. First locate the exact break, make a visible attempt, and ask for the smallest hint that lets you continue. Use more help only when the smaller step is not enough.

This approach preserves the useful part of struggling: making decisions, testing ideas, and noticing what you do not yet understand. It also prevents unproductive struggle from becoming an hour of repeating the same failed move.

What does “stuck” actually mean?

Before looking for help, finish this sentence:

I can get as far as ___, but I cannot decide/remember/explain ___ because ___.

You may discover that you are stuck in one of several places:

• you do not understand a word, symbol, instruction, or condition;
• you cannot recall the relevant idea;
• you do not know which method applies;
• you can choose the method but cannot start;
• you made progress and cannot justify the next step;
• you reached an answer that conflicts with another fact;
• you can solve this version but not a changed one.

“I don't get any of it” may be how the problem feels, but it gives you nothing to act on. Naming the last point that still makes sense turns a wall into a location.

What is the help ladder?

Move up this ladder one rung at a time. After each rung, return to the problem before asking for more.

1. Restate the problem

Say what the task is asking in ordinary language. List what you know, what you need, and any constraints.

For writing, identify the decision the paragraph must support. For code, state the expected input and output. For a language task, say what meaning you are trying to express. For a calculation, name the quantities and relationship.

If you cannot restate the task, your first need is clarification—not a solution.

2. Show one attempt

Draw, write, speak, calculate, outline, or test something. A wrong attempt is more informative than a blank page because it reveals your current model.

Ask:

• What approach did I try?
• What did I expect to happen?
• What happened instead?
• Which part still seems valid?

3. Retrieve before you search

Without opening your notes, write the ideas that might matter. Then check your materials and correct the list.

This separates “I forgot the tool” from “I have the tool but cannot apply it.” The Open University recommends recalling first rather than relying on rereading.

4. Ask for one cue

A cue points your attention without choosing every step. Useful cues include:

• Which condition have I not used yet?
• What kind of problem is this similar to?
• Which term or concept should I revisit?
• Is my setup valid so far?
• What should I draw or compare?

If the cue is enough, continue alone.

5. Ask for the smallest useful hint

A hint should change your next action without completing the task. Try:

Do not solve this for me. Look at my attempt, identify the first wrong or missing decision, and give me one hint for what to try next.

If you are using AI, include your course rules and the relevant source material. Generated explanations can be wrong, and a helpful-looking answer may still cross an academic-integrity boundary set by your instructor or organization.

6. Reveal one worked step

If the hint is not enough, inspect only the first step you cannot produce. Cover the rest and try to continue.

The University of Washington's study guidance recommends looking at one step of a worked solution, then giving yourself another attempt before revealing more.

7. Study the full solution—then close it

Sometimes you need the complete worked example. Study the choices, not only the final answer:

• Why was this approach selected?
• Which clues mattered?
• Where did it differ from your attempt?
• What would change in a similar problem?

Then close the solution and reconstruct it. Later, solve a fresh version from a cold start. Otherwise the solution may create the feeling of understanding without the ability to produce it.

How long should you struggle before asking for help?

There is no universal number. Use progress rather than a timer as the main rule.

Keep working while you are producing new representations, testing new ideas, or narrowing the uncertainty. Ask for help when you are repeating the same move, guessing without feedback, or no longer know what evidence would change your mind.

A short break can help when attention is exhausted. It cannot replace missing knowledge, feedback, or a prerequisite concept.

What should you write down before asking someone?

Give a teacher, tutor, classmate, or AI four things:

1. The task: what you are trying to do.
2. Your interpretation: what you think the question means.
3. Your attempt: the work you have done so far.
4. The break: the first decision or step you cannot make.

For example:

I need to compare these two arguments. I think the first uses stronger evidence, and I have identified both claims. I get stuck deciding whether the second source is less credible or simply supports a narrower claim. Can you ask me a question that helps me distinguish those?

That is much easier to teach from than “What is the answer?”

How do you know the help worked?

Help worked when you can continue with less help—not when the completed answer looks clear.

After receiving help:

1. explain why the hint or step was useful;
2. finish the current problem yourself;
3. try a fresh problem without the hint visible;
4. return later and check whether you can still choose the approach.

If you can follow the solution but cannot start the next one, read why an explanation can make sense while independent solving still fails.

The goal is not to avoid help. It is to use help that gives the thinking back to you.

Talk the stuck point through with Kuji, see it take shape on a shared whiteboard, and keep the next move in your hands.