The curse of knowledge is simple: once you know something, you cannot simulate not knowing it. So your writing silently assumes context the reader does not have, and the gaps are invisible to you, because to you there are no gaps.
This is why the smartest person on the team frequently writes the most confusing messages. Carelessness and arrogance have little to do with it. Expertise removes your access to the beginner's view, the way you cannot unsee the arrow in the FedEx logo once someone points it out.
How the curse looks on the page
The curse has recognizable fingerprints, easier to spot in others' writing than in yours, which is the whole problem:
- Unexpanded shorthand: acronyms, project codenames, "the Q2 thing," names dropped as if universally known.
- Skipped steps: instructions that jump from step two to step five because three and four became automatic for you years ago.
- "Obviously" and "just": the words that mark exactly where the author stopped being able to see the difficulty. "Just redeploy the service" is a sentence written by someone who no longer remembers learning to deploy the service.
- Conclusions without the ladder: you climbed four rungs of reasoning to your position, then wrote only the top rung.
That last one causes the most damage in decision-making threads. To you, "we should switch vendors" is the obvious summit of a mountain of evidence. To the reader who never saw the mountain, it is an opinion floating in space, and they respond to it as opinion, and you get frustrated that they "don't get it." Nobody in this exchange did anything wrong. The evidence just never made it out of your head.
You cannot fix it by trying harder
The standard advice, "put yourself in the reader's shoes," fails against this particular bias, and it is worth being blunt about why: the shoes are gone. Your pre-knowledge state is not stored anywhere you can visit. Effort and empathy do not bring it back.
What works instead is mechanical, outside-the-head checks. The reliable ones:
- The new-reader test: someone smart but uninvolved reads it and marks every place they stumble. Cheapest quality control in existence, used least.
- The question forecast: before sending, write down the first question the reader will reply with. If you can predict it, answer it in the message. You can almost always predict it.
- The glossary sweep: reread only for nouns. Every acronym, codename, and internal shorthand either gets one clause of expansion or an on-purpose decision that this reader knows it.
- The steps count: for instructions, do the task while reading your own steps, literally. Step three-and-a-half always falls out.
Writing Unvent's own onboarding drilled this into me. A line I once wrote said to press the shortcut "as usual," and to me it truly was usual; I had pressed it several thousand times. Watching where new users actually stopped taught me the sentence was written for an audience of one, and that audience had built the thing. The fix was writing down the two steps my hands had stopped counting.
Jargon is a special case: sometimes the curse, sometimes a key
The curse of knowledge conversation usually collapses into "avoid jargon," and that is too crude. Between two specialists, jargon is compression: "idempotent" saves a paragraph. The curse is losing track of which reader holds which keys.
So the honest rule is per-reader, not per-word. Writing to one colleague who shares your vocabulary, compress away. Writing to a mixed thread, expand once at first use and then compress. Writing to a client outside your field entirely, translate to effects and drop the mechanism. The failure is running the same vocabulary across audiences on autopilot, because the autopilot is exactly the part of you the curse controls.
The curse gets worse as you get better
Here is the uncomfortable trajectory: every year you get more expert, your writing to non-experts gets naturally worse, unless you build counter-habits. Expertise means more of your knowledge going automatic, and automatic knowledge is precisely the knowledge you can no longer see yourself using.
This is why senior explainers are rare and disproportionately valued. It is also why "they're too smart to explain things" is backwards: the explaining failure is an unmaintained bridge. The people who stay good at it are running maintenance: teaching juniors regularly, keeping one non-expert friend who asks real questions, rereading their own old beginner-era messages, and never trusting the feeling of "this is clear," because that feeling reports on the writer's head, not the reader's.
The practical takeaway fits in a pocket. The feeling of clarity is not evidence of clarity. Evidence comes from outside: a test reader, a predicted question answered, a step list walked through with your own hands. And on the sentence level, when a draft has curled up into shorthand, a plain-language rewrite where you are typing is a fast way to see your own message the way a stranger will, which is the one perspective your expertise took from you.
Write for the reader you cannot remember being
The clear version of your message is obvious to everyone except its author. Unvent rewrites what you typed, in place, so the reader gets the steps your head skipped.
Add Unvent to Chrome