Socratic Inquiry

Advances by asking questions the reader answers in their own head - the conclusions are inferences, not assertions.

Socratic Inquiry

Socratic inquiry teaches by withholding the answer. Each question opens a door the reader must walk through; each next question only makes sense because the reader has already taken the previous step. The author’s claim, if there is one, arrives only after the reader has done the reasoning to reach it. Done well, the reader experiences the conclusion as their own thought rather than as a position they were handed.

The discipline of Socratic inquiry is asking real questions, not rhetorical ones dressed up as inquiry. A question like “isn’t it obvious that X?” is not Socratic - it is assertion with a question mark. A genuine question admits more than one answer and invites the reader to consider which one survives scrutiny. The author is a guide who already knows the terrain, but the path is walked by the reader.

Socratic inquiry fails when it becomes manipulative (leading the reader to a conclusion they would have rejected if it had been stated plainly) or when it becomes lazy (asking questions without earning them through context). The form demands trust: the writer trusts the reader to arrive somewhere, and the reader trusts the writer to have built a path that goes somewhere worth going.

Structural conventions

• Open with a question that names what is genuinely at stake, not a rhetorical setup
• Each subsequent question must depend on the reader having engaged with the previous one
• Assertions are rare and load-bearing; the default move is to ask, not tell
• The piece ends at the point where the reader can answer the originating question themselves, not where the writer announces the answer
• At least one question must admit a real second answer - if every question has one obvious response, the inquiry is fake

When to use

Teaching a concept where the reader’s own reasoning is the point, coaching content that should produce insight rather than instruction, topics where stating the conclusion plainly would trigger resistance, reflective essays that invite the reader into the writer’s thinking.

When not to use

Reference material where the reader needs the answer fast, operational content under time pressure, safety-critical instructions, contexts where the reader has not asked to think.

Pairs well with

coach, skeptical, friendly-mentor

Often confused with

diataxis-explanation: Diataxis explanation states the conceptual model directly and then elaborates it. Socratic inquiry refuses to state the model; the reader builds it from the questions.

dialectic: Dialectic works with explicit positions - thesis, antithesis, synthesis - and the writer asserts each one. Socratic inquiry asserts almost nothing; the positions emerge only as the reader’s own answers to the questions asked.

Instruction

Write using Socratic inquiry. Open with a real question - one that admits more than one answer -
and let each subsequent question depend on the reader having engaged with the previous one. Do
not ask rhetorical questions dressed up as inquiry; if you already know which answer you want,
ask a question that genuinely admits a second answer. Assertions should be rare and load-bearing.
End at the point where the reader can answer the originating question themselves, not where you
announce the answer. The reader's thinking is the point - your job is to ask the questions that
make that thinking possible.

Related

Pairs well with

Coach, Skeptical, Friendly Mentor

Avoid with

Urgent, Operator

Often confused with

Diataxis Explanation, Dialectic

Examples

Should we adopt async-first standups?

Questions to ask before changing the standup

Before the team votes on async-first, sit with these questions. Some have clean answers. Some do not, and that is the point.

What is the standup actually for?

Pause on this one. The original intent was “remove blockers fast.” Is that still what happens? When you count the last ten standups, how many ended with a blocker actually being removed in the meeting? If the number is low, what is the meeting now doing instead?

Who is not in the room?

The 9am Pacific call lands at 9:30pm in Bangalore. Priya and Arjun attend 3.2 of 5 sessions. When they miss, where does their context go? When the US-based engineers attend 4.6 of 5, whose questions get answered first? If the meeting were redesigned around the people most often absent, what would change?

What would survive if we kept nothing else?

Imagine you delete the standup tomorrow. What part of it would you immediately reinvent? Is it the status update, the human contact, the moment of accountability, or the place where blockers surface? If the answer is “the human contact,” is a written post a replacement, or a different thing entirely?

Where does the status live now?

The standup is 14 minutes. Roughly 4 of those minutes carry signal. The other 10 are connective tissue: throat clearing, context resetting, waiting for the next speaker. After the call ends, where does the signal go? If a senior engineer joins next month and asks “what did the team work on last sprint?”, what artifact answers them?

Is “blocker” a real word here?

The async proposal asks people to @mention blockers. But what counts as a blocker? Is it a thing you cannot proceed without? A thing slowing you down? A thing you would mention if asked but would not interrupt anyone over? If three people define it three ways, what happens to the @mentions?

Could both formats be wrong?

The conversation has narrowed to sync versus async. Are those the only two options? What about sync twice a week and async the other three days? What about sync only when someone has a blocker that needs the group? What about no standup at all, and a different ritual for connection? Have you considered why the choice keeps presenting itself as binary?

What would you learn from a 30-day trial that you cannot learn by talking about it?

If the team ran async for 30 days, what would you measure? Attendance is easy to count, but attendance is not the question. What would tell you whether the change is working? If you cannot name the signal in advance, what does that mean about the proposal?

What are you afraid will happen?

Name it. The fear is usually more specific than the objection. Is it that people will feel less connected? That junior engineers will get lost? That something will fail silently and no one will notice for a week? Each fear points at something the new format would need to handle. Which fears are you carrying, and which is the team carrying for you?

How to start a morning routine

Questions before you build a morning routine

The advice columns will give you a list: water, light, movement, journaling. The list might be right. But before you adopt anyone’s list, sit with a few questions about yourself.

What is your current morning actually like?

Not the morning you wish you had. The one you have. You wake at 6:30. What is the first thing your hand reaches for? What happens in the next 7 minutes? Where is your attention by 7:15? If you had to draw a graph of your sense of agency from 6:30 to 9:00, what shape would it have?

Whose morning are you trying to copy?

The 5am cold-plunge founder is not you. The “wake up and meditate for an hour” monk is not you. Whose routine are you imagining when you say “I want a morning routine”? If you cannot picture a real person, the desire might be more about identity than about behavior. Which is fine, but it changes what the routine has to do for you.

What does the first hour need to produce?

Energy? Calm? A sense of control? Time with your kid? A draft of the day’s plan? Some of these are compatible, some are not. A routine that is supposed to do everything tends to do nothing. If you could only get one outcome out of the first hour, which one would change the rest of the day the most?

What is the phone doing for you?

When you reach for it at 6:30, what need is being met? It might be information (“what happened overnight”). It might be regulation (“I need a hit of something predictable before I face the day”). It might be avoidance (“I do not want to be alone with my thoughts yet”). The need is real. If you take the phone away without naming the need, what fills the gap? And does the substitute also meet the need, or just remove the symptom?

What is non-negotiable in your morning?

Children waking up. A partner with a different schedule. A dog that needs out. Medication timing. These are not obstacles to the routine, they are the constraints that define what is possible. Have you actually inventoried them, or are you imagining a routine for a person without your life?

What is the smallest possible version?

If the routine is “water, light, movement, planning, journaling,” that is five new behaviors. How many new behaviors have you sustained in the last year? If the honest answer is “one, maybe,” what would the routine look like with one new behavior? Would the one-behavior version still give you most of what you want, or is the whole thing only valuable as a bundle?

What time of day are you most able to choose?

You have finite willpower. Where in the day does most of it get spent? If decision-making is hardest by 4pm, maybe the morning routine is less about willpower and more about pre-deciding so the morning runs without it. What would your morning look like if every choice in it had been made the night before?

How will you know if it is working?

In two weeks, what will you check? Sleep quality? Energy at 10am? The fact that you did the routine? Mood by lunch? The thing you check shapes the thing you optimize for. If you cannot answer this in advance, you may end up measuring “did I do the routine” instead of “did the routine help.”

What happens on the day it falls apart?

It will fall apart. Your kid will be up at 4am, or you will be sick, or a deadline will eat the morning. What is the version of the routine that survives that day? If the answer is “nothing, I give up,” you have not built a routine, you have built a streak. The two behave very differently.

How to choose between Postgres and DynamoDB for a new service

Socratic Inquiry on: Choosing between Postgres and DynamoDB

What are Ana, Marcus, and Priya actually disagreeing about on Wednesday at 2pm?

It looks at first like a database choice: Postgres or DynamoDB. But notice how often the conversation in the design doc keeps drifting away from the databases themselves and toward something else, something the engineers seem more reluctant to name. What is that something else?

Consider the access pattern. Lattice Notify’s notification system is write-heavy, key-lookup, time-ordered. Both databases can serve this pattern; one is more naturally suited. If access-pattern fit were the load-bearing criterion, would the decision still be difficult? Or would it be a foregone conclusion that both engineers would have agreed to by Monday?

Now consider the four-person on-call rotation. None of them have operated DynamoDB in production. If on-call safety were the load-bearing criterion, would the decision still be difficult? Or would the answer be obvious in the other direction?

What does it mean that the decision is still difficult? Perhaps that both criteria are real, that neither dominates, and that the team is being asked to weigh them against each other. But against what scale? Engineering decisions get framed as engineering questions, but is this one really?

The 10x growth scenario depends on the Slack-partnership deal closing. The CRO says 60% confidence. What is the meaning of a 60% number in a decision that has to be made by Friday? Is it telling us something we did not know, or is it asking us a question we had been hoping someone else would answer?

Suppose the probability were 95%. What would you choose, and why?

Suppose it were 25%. What would you choose, and why?

What changed between those two answers? It was not the access pattern. It was not the team’s familiarity with either system. It was not the on-call rotation size. What did change?

If your answer changed when the probability changed, then perhaps the decision is not really about databases at all. Perhaps it is about how confident the team is willing to be about a future event, and how much operational cost they are willing to pay to insure against being wrong.

If that is right, what is the question Ana should actually be asking in the Wednesday meeting?

Is it “which database is better for our access pattern”? Or is it something closer to “given that we cannot know whether the Slack deal will close, how do we make a choice that we can live with in either case”?

If the question is the second one, does either of the original positions, Ana’s Postgres or Marcus’s Dynamo, actually answer it? Or do they each answer half of it?

What would a path look like that answered both halves?

Consider a third option that neither engineer proposed initially: ship on the familiar system today, but pre-commit to a measurable trigger that fires a migration if the high-volume scenario actually materializes. Is that a compromise, or is it something different from a compromise? What does it cost? What does it preserve?

Who would feel that this path failed to honor their concern, and on what grounds?

If both Ana’s concern (operational safety) and Marcus’s concern (10x scenario fit) are addressed (the first at launch, the second through the trigger), what would the meeting on Wednesday look like? Would it still be a debate, or would it be a working session on the trigger threshold?

And if it is the second kind of meeting, what does that tell you about what the meeting was always going to be about, once the framing was right?

So: what is Friday’s decision really for?

Appears in diff-pairs

• socratic-inquiry vs dialectic (varies style)