The Meta-Skill of Questioning

Asking a good question is becoming more important than getting a good answer.

A well-framed question doesn’t just elicit an answer—it sparks deeper thinking, invites insight, and unlocks disproportionate value. Yet asking well remains surprisingly hard.

Poor questions tend to share these traits:

Too broad: “What do you think about this?” or “Any suggestions?”
Common-knowledge queries: “What is GEO?”
Overly verbose: A five-minute preamble before the actual question emerges.
Self-answering: Framing a question but immediately supplying the full answer—blurring the line between inquiry and declaration.

Even more common than poor questioning is avoiding questions altogether.

I wrote earlier about this in “Asking Too Little Costs You”:
mp.weixin.qq.com

That piece scratched the surface—but what lies beneath technique is something deeper: the meta-skill of questioning.

Think of it this way: The meta-skill of questioning is the ability to reflect on, monitor, and iteratively improve the act of questioning itself.

It’s not whether you can ask—but how consciously you shape your questions, how deliberately you calibrate them for clarity, relevance, and leverage—and how persistently you refine that process over time.

To train this meta-skill, try these three reflective prompts before asking anything:

What is my core confusion right now?
What background does the other person need to understand it?
Can I compress the whole question into under 100 words?

From those, a practical framing emerges:

In [context], if [assumption], then how does [key variable] change?

So—how do we actually build this meta-skill? At minimum, three interlocking abilities must be cultivated:

Listening: At Yuanfudao, one core value is “Reflect, Listen, Simplify, Iterate”—and “Listen” sits first. As one leader put it: “Leadership is listening to the whole system better than anyone else.” I’ve met hundreds of people—and most lack listening awareness. They have expression impulse, not listening discipline. That’s why their questions stay shallow.
Thinking: Pause for just three seconds before speaking or typing. That tiny gap lets you digest what you’ve heard, reinterpret it, and reframe your question—not just react.
Expressing: Ultimately, questioning is an act of communication. Time-tested tools still hold up: the Pyramid Principle, the Golden Circle, and Nonviolent Communication—all sharpen precision, hierarchy, and empathy in delivery.

Now it’s clearer why asking well feels so rare. It demands integration—not just one skill, but three, practiced together.

Integrating Lived Experience

Over the past few years of entrepreneurship, I’ve met many founders who’ve delivered outsized results. In our conversations, certain ideas they repeatedly emphasized—sometimes almost reflexively—have proven deeply valuable in hindsight.

Examples include:

“Running a company means managing risk—not revenue.”
“Profit follows reputation.”
“Video account ads have real ROI—target by audience package, and always test with short videos first.”
“The book title matters most—spend real time designing it.”
“If a robot can do it, don’t assign it to a human.”
…

The signal here isn’t just repetition—it’s resonance across time and context. Yes, path dependence and subjectivity color each person’s emphasis—but what survives repeated telling is distilled, battle-tested insight.

History works the same way. Centuries of conflict, governance, and survival collapse into terse maxims: “Know thyself.” “Power corrupts.” “Divide and rule.”

I once spoke at length with a mega-bestselling author about reading habits. His single strongest recommendation? “Read history—deeply and often.”

Mao Zedong was a voracious reader of Chinese history. His strategic innovations drew directly from historical patterns:

“When the enemy advances, we retreat; when they halt, we harass; when they tire, we strike; when they withdraw, we pursue”—echoes of peasant uprisings against imperial forces.
“On Protracted War”—informed by centuries of stalemate warfare against northern nomads and coastal pirates.

But raw experience—historical or personal—doesn’t automatically become wisdom. To truly integrate it, we must reflect, connect, and apply: test insights against new situations, revise assumptions, and let practice deepen understanding.

Effective Information Volume

Prompt engineering alone often falls short of what modern AI agents need to perform robustly. That’s why “context engineering” is gaining traction.

During a live stream, I asked viewers: What’s the core difference among prompt, prompt engineering, and context engineering? The question sparked real debate.

Here’s how to distinguish them simply:

Prompt: A single instruction given to an AI.
Prompt engineering: A systematic process—designing, testing, iterating—to build reliable, reusable prompts. Like software engineering, it’s iterative and outcome-oriented.
Context engineering: Goes beyond the prompt. It layers in background context, reference materials, user profile data, and even the AI’s memory of prior interactions with you.

At their root, the key differentiator is effective information volume—not total data, but signal-rich, task-relevant information. Context engineering delivers more of it—so outcomes improve.

The progression is clear:
Prompt → Prompt engineering → Context engineering
…a natural evolution from simple to complex, static to adaptive.

A few implications:

More data ≠ better output. Effective information does. Example: Instead of listing 20 web design specs, tell the AI: “Analyze the user’s input, infer content type, find the world’s best-performing comparable page, then generate a matching design.”
Understanding “effective information” transforms prompt writing—it becomes less about phrasing and more about what information to prioritize and how to structure it for the AI’s reasoning.
Different business needs demand different levels: A customer service bot may thrive on strong prompt engineering; a personalized learning tutor requires full context engineering.

This lens applies far beyond AI—it sharpens how we brief colleagues, write emails, or plan projects.

Four Principles for AI-Native Childhood Development

AI isn’t just a tool—it’s a high-fidelity medium for parent-child co-creation. Over recent years, my daughter and I have built dozens of shared experiences using AI: illustrated storybooks, AI-generated videos, custom agents, AI-composed lullabies—even voice-cloned digital twins.

These aren’t just fun diversions. They’re low-stakes, high-engagement training grounds for future-ready thinking. Here’s what we’ve learned:

1. Precision in Expression

We’ve generated hundreds of images together using ChatGPT, Jimo (Ji Meng), and Midjourney—all driven by her spoken descriptions.

Our ritual: She names features (“a princess wearing a crown, sparkly shoes, a necklace, and a long dress that touches the floor”). I type exactly what she says—no editing—then feed it to each model. She picks her favorite. I paste it into Keynote, print in color, and she cuts it out with scissors. Each piece is authentically hers.

Early on, she’d point and say, “I want this one!” To stretch her precision, I’d gently block her hand and say, “I don’t understand ‘this one.’ Tell me: row two, third image?”

She quickly internalized the habit—shifting from vague pointing to structured coordinates. In the AI era, precision isn’t pedantic—it’s power.

2. No Boundaries, No Judgement

We never label her ideas “right/wrong” or “good/bad”—whether sketching aliens or scripting AI video plots. It’s tempting to steer (“Let’s make the robot kinder!”), but restraint is essential.

Why? Because unfiltered imagination is where true creativity lives. I catch myself constantly—and mostly succeed—in pausing before correcting. The payoff? Her ideas grow bolder, weirder, and more original.

3. Knowing Thyself (Through Digital Mirrors)

We created voice-cloned digital twins—“Yao Rui 2,” “Yao Rui 3,” etc.—and she regularly chats with them on my phone. She draws characters, assigns them personalities, and stages debates between them.

This isn’t playacting. It’s early scaffolding for self-dialogue—a hallmark of mature cognition. Adults who thrive long-term are often those who’ve learned to host thoughtful inner conversations.

4. Delayed Gratification, Designed

We once made an empty chocolate box, sealed it, and told her: “Open it in seven days—you’ll find exquisite chocolate inside.” She counted down daily.

With AI, we extend this: co-designing simple games where rewards unlock only after debugging, refining, or waiting. The process of building and tuning the game—not just the final output—builds patience, iteration stamina, and intrinsic motivation.

Cosmic-Perspective Meditation

Try this during meditation: zoom your awareness outward—past Earth’s atmosphere, past the Moon, past Mars and Jupiter—until you’re floating in deep space, watching our pale blue dot rotate silently in the void.

This is “cosmic contemplation” (guānxiǎng)—a form of universe-scale contemplation. Its effects are quietly profound:

Psychological distance: From orbit, daily stresses shrink to near-nothing. That deliberate zoom-out lowers emotional intensity and helps the brain exit fight-or-flight—making space for calm, systems-level reflection.
Existential belonging: Feeling kinship with stardust fosters deep, quiet grounding—not “I am small,” but “I am of this.”
Awe as cognitive reset: Awe dissolves rigid self-world boundaries. It opens mental bandwidth—making room for curiosity, humility, and unexpected connections.
Time recalibration: Fifteen minutes of cosmic meditation can feel like hours—long enough to soften the grip of past regrets and future anxieties. In that expanded “deep time,” presence becomes effortless.