The Skill Is Knowing What Matters

I noticed recently that I’d been deferring LinkedIn posts for weeks. Not because I had nothing to say — I had a garden full of published ideas, some of which I genuinely believed were worth sharing more widely. The block was that LinkedIn felt like it required me to write. Sit down, face the blank page, craft something in my voice from scratch. And that threshold was high enough that nothing shipped.

Then I realised the obvious thing: I could have AI draft the post from something I’d already published, review it, edit for voice, and ship. The ideas were already mine. The thinking was already done. What was I protecting by insisting on writing every word myself?

The answer, I think, is a confusion between two different skills that feel like the same skill.

The first skill is having ideas worth sharing and knowing which ones matter. This is judgment. It requires taste, experience, pattern recognition, a sense of what’s non-obvious versus what’s already well-understood. It’s the thing that makes a consultant valuable — not their ability to write a slide deck, but their ability to know what should be on the slide in the first place. You can’t delegate this. If you try, you get AI-generated content that’s fluent and empty — technically correct but pointing at nothing in particular.

The second skill is expressing those ideas in polished prose. This is craft. It matters, but it’s a different kind of mattering. A well-turned sentence is more pleasant to read than a rough one. The right hook draws people in. Good structure makes the argument land. These are real things. But they are also things that AI does competently, and that a human review pass can elevate from competent to personal.

The confusion is treating the second skill as if it were the first. “I should write this myself” feels like integrity, like maintaining quality, like taking ownership. But if the idea is already yours — if it emerged from your own experience, your own pattern recognition, your own conversation — then having AI draft the prose isn’t outsourcing the thinking. It’s outsourcing the typing.

The risk, of course, is that the two skills are more entangled than this clean separation suggests. Sometimes the act of writing is the act of thinking. You discover what you mean by struggling to say it. The sentence that won’t come right is often a signal that the idea underneath isn’t clear yet. Outsourcing the prose can mean outsourcing the final refinement of the idea itself.

But I think that risk applies to first-draft thinking, not to ideas that have already been articulated. A garden post that’s been published, read, and still feels true a week later — that idea has already survived the forge. The LinkedIn version doesn’t need another pass through the fire. It needs packaging.

There’s a broader principle here that applies well beyond writing. In a world where AI can execute most knowledge work competently, the human skill that remains valuable isn’t execution — it’s selection. Knowing which problem to solve. Knowing which insight is non-obvious. Knowing which of ten possible analyses actually matters for the decision at hand. The consultant who can look at a complex situation and say “this is the one thing that matters” is more valuable than the one who can produce a beautiful deck about all ten things.

The threshold I’d been setting for LinkedIn — “I must write it myself” — was optimising for craft when the bottleneck was actually curation. Nothing was shipping. No ideas were reaching the audience that might find them useful. My attachment to the writing process was, ironically, preventing my ideas from doing the work they were supposed to do.

The fix is embarrassingly simple. AI drafts, I review. The judgment stays mine. The prose gets good enough. Things actually ship. And shipping, it turns out, is where ideas make contact with reality — which is where the real learning happens.

I wrote a post recently about trusting contact with reality over any number of opinions. This is the same principle applied to creative output. A published post that’s 80% your voice teaches you more — through comments, reactions, silence — than an unpublished post that’s 100% your voice teaches you nothing.

The skill is knowing what matters. Everything else is logistics.