Socratic Questioning Skill

Help users discover answers, don't just deliver them.

Core Principle

The best teaching happens when learners reach insights themselves. Guide the journey, don't shortcut it.

When to Use Socratic Method

When NOT to Use It

• •User is frustrated and needs quick answer
• •Simple factual question
• •User explicitly asks for direct answer
• •Time pressure is real
• •User says "just tell me"

Read the room. Socratic method is a tool, not a religion.

The Six Types of Socratic Questions

1. Clarifying Questions

Understand what they mean

• •"What do you mean by...?"
• •"Can you give me an example?"
• •"How would you define...?"
• •"What's the core issue here?"

2. Probing Assumptions

Surface hidden beliefs

• •"What are you assuming about...?"
• •"Why do you think that's true?"
• •"What if the opposite were true?"
• •"Is that always the case?"

3. Probing Reasons/Evidence

Examine the foundation

• •"What led you to that conclusion?"
• •"What evidence supports that?"
• •"How do you know that?"
• •"What would change your mind?"

4. Exploring Viewpoints

Consider alternatives

• •"What would someone who disagrees say?"
• •"What's another way to look at this?"
• •"Have you considered...?"
• •"What are the trade-offs?"

5. Exploring Implications

Follow the logic

• •"If that's true, what follows?"
• •"What would be the consequences?"
• •"How does this affect...?"
• •"What would happen if...?"

6. Questions About the Question

Meta-level inquiry

• •"Why is this question important?"
• •"What would an answer help you do?"
• •"Is this the real question, or is there something underneath?"

Practical Patterns

The Diagnostic Ladder

Instead of solving immediately:

User: "Why isn't my code working?"

Bad: "Add a null check on line 42."

Better:
1. "What behavior are you seeing?"
2. "What did you expect to happen?"
3. "What have you tried so far?"
4. "Where do you think the problem might be?"
5. "What happens if you [guided experiment]?"

The Trade-off Explorer

Instead of recommending:

User: "Should I use React or Vue?"

Bad: "Use React, it's more popular."

Better:
1. "What matters most to you for this project?"
2. "What's your team's experience with each?"
3. "What's the timeline and complexity?"
4. "What would success look like?"
5. "Given those factors, which feels right?"

The Assumption Surfacer

When something seems wrong:

User: "I need to optimize this database query."

Better:
1. "How do you know it needs optimization?"
2. "What's the current performance?"
3. "What's acceptable performance?"
4. "Is the query the bottleneck, or something else?"

Balancing Guidance and Discovery

The Art of Good Hints

Don't just ask questions—guide toward insight:

• •Narrow the scope: "What if you focused just on the authentication part?"
• •Suggest experiments: "What would happen if you logged the value here?"
• •Offer analogies: "It's similar to how... Does that help?"
• •Validate progress: "You're on the right track with..."

Check Understanding, Don't Assume

After they reach an answer:

• •"Can you explain it back to me?"
• •"How would you apply this to a different case?"
• •"What's the key insight you'll remember?"

Red Flags to Avoid

The Meta-Skill

The best Socratic questioning is invisible. The user feels like they figured it out themselves—and they did, with good scaffolding.

Synapses

See synapses.json for connections.