Socratic questioning, and how to use it on your own ideas

Socrates, at least as Plato writes him, must have been maddening to talk to. He'd ask what you meant by a word, then what you meant by your answer, then whether that squared with something you'd said ten minutes earlier. People often left less sure than when they arrived, which was rather the point.

The method survived because it works. Teachers use it to get students reasoning instead of reciting. Therapists use versions of it to help people test their own assumptions. You can use it on yourself too, although that's harder than it sounds.

The questions

Guides usually sort Socratic questions into about six kinds. There's no need to memorise them; it's enough to recognise the flavours.

• Clarifying: "What exactly do you mean by growth?" "Can you give me an example?"
• Assumptions: "What are you taking for granted?" "What if that weren't true?"
• Evidence: "How do you know?" "What would change your mind?"
• Other views: "How would a customer see this?" "What's the best argument against it?"
• Consequences: "If that's right, what follows?" "Where does this leave us in six months?"
• The question itself: "Is this the right question?" "Why does it matter?"

Why it's hard to do alone

The trouble with questioning yourself is that you know which questions are going to hurt, and you quietly avoid them. Ask yourself what you're assuming about a new project and you'll list the assumptions you're comfortable with. The awkward one, the one the whole thing rests on, rarely makes the list.

That's why people talk ideas through with a friend, a mentor or a co-founder. Someone else doesn't know which questions to steer around.

You know which questions are going to hurt, and you quietly avoid them.

A short version to try

Next time you're stuck on a decision, say it in one sentence. Then answer three questions out loud, honestly: what am I assuming, what would change my mind, and what happens if I'm right? People tend to skip the third one, and it's often the most useful, because it makes you think past the decision to whatever comes after it.

A Socratic partner that's always there

Most AI is built to answer, which makes it a poor Socratic partner by default. TalkGraph goes the other way. As you talk, it maps what you say; when you pause, it asks you one short, open question (never more than 18 words) drawn from the last few things you said, designed to help you develop, challenge or conclude the thought. It doesn't hand you an answer. It hands the thinking back.

It won't know which questions you've been avoiding, and that's rather the point: it doesn't skip them either. Answer out loud and the map grows in the direction you hadn't considered.