The Myth of “Just Make Great Work”

Somewhere right now, a piece of work worse than yours is winning. Not on merit. On whether a machine could confirm who made it.

The Cost You Don’t See Yet

Every week, someone asks an AI assistant a plain question inside your field. The assistant answers with a name. Most weeks, for most people reading this, that name isn’t theirs.

The work is not thin. The assistant has nothing solid to attach to a name — no consistent identity, no confirmed track record, no independent source backing up who this person is and what they’ve built. So it reaches for whoever it can confirm, even when that person’s work is thinner than yours.

This is not a one-time loss. It repeats every time the question gets asked again, and the questions never stop. Every week you stay unattached to your own work is a week someone else’s average output quietly banks the citation that should have gone to yours.

This is not a single missed introduction you can shrug off. It is the same loss, repeating on a schedule you don’t control, every time someone in your field opens an assistant and asks. A year of that is not a rounding error. It is a year of citations accumulating somewhere else, attached to a track record that does not match yours.

Why Machines Don’t Grade on a Curve

Name the mechanism plainly: entity consistency.

A system assembling an answer is not reading your report and weighing its rigor. It has no taste to apply. What it does instead is try to resolve a stable, confirmable identity behind a piece of work, then check whether independent sources back that identity’s authority on the subject.

If your name, your one-line description of who you are, and your track record show up differently everywhere your work appears — or don’t show up at all — the system has no single identity to resolve. It has scattered fragments it cannot confidently stitch together. Fragments don’t get cited. Confirmed identities do.

This is why confidence so often beats correctness inside these systems, and it isn’t a design flaw worth complaining about. It’s the plain mechanics of attribution at scale. A name that resolves cleanly, corroborated by sources that agree with each other, gets treated as more citable than an unattached claim — regardless of which one is more rigorous underneath.

Corroboration matters as much as consistency. A system checking whether to trust a claim about you looks for more than one place saying the same thing, not only the strongest single sentence you wrote about yourself. Freshness matters too. A claim that was true and checkable two years ago, and hasn’t been touched since, reads differently than one connected to something recent and still standing.

Notice what this does to the whole question. The fix has nothing to do with making the work better. The work can already be excellent. What’s missing is the layer around it: the part that lets a stranger, or a machine, confirm in seconds who is responsible for it, and whether that claim holds up anywhere else.

The work speaks. But someone has to open the window so the world can hear it — and the window, in plain technical terms, is exactly this: a consistent, confirmable identity attached to everything you publish, corroborated by more than your own say-so.

What Goes Where, First

Start with the piece of work itself, not your website. Attach your name and one factual sentence about who you are directly to it, at the top, not three clicks away in a footer nobody opens. That sentence is the anchor a machine needs in order to answer “who made this” — and most people either bury it or skip it, the same way an excellent report can say nothing at all about who wrote it.

Keep that sentence identical everywhere it appears. Your site’s about page, a guest byline, a directory listing, a speaker bio. If your title, your years, or your claim shift even slightly from one to the next, you stop being one confirmed identity. You become several half-matching fragments, and fragments don’t get resolved with confidence.

Publish one canonical bio page: plain, factual, crawlable, stating who you are, what you’ve built, and the question you answer. Then check it the unglamorous way — fetch that page with scripts turned off and confirm the text is sitting in the raw response. Credentials that only render after a script runs are, to a simpler crawler or agent grabbing the page, credentials that were never published at all.

This matters more than it sounds like it should. A growing share of the systems reading your page do not wait for every script to finish before deciding what the page says. They take the fastest version they can get. If your identity only appears after something loads, the fastest version of your page — the one many systems see — is silent about who you are.

Making It Hold Up

Where the platform allows it, attach structured author information to the piece — the standard, widely supported markup that lets a page declare, in a form a machine parses cleanly, who wrote it. This isn’t decoration. It’s the difference between a name a human can read and a name a machine can extract and trust.

Revisit that structured information whenever your track record changes: a new credential, a new piece of work worth naming. Stale structured data is barely better than none. It confidently tells a system you are what you used to be, not what you’ve since become.

Then go get corroboration. One independent mention — a colleague citing you, a directory listing your record, an outside source naming you in connection with the work — does more for your resolvability than another paragraph you write about yourself. A claim only you have made is weak evidence. The same claim echoed independently is a fact worth repeating.

One independent analyst once wrote a report tighter than the firm that beat her to the client — extra verification included, unbilled, on her own time. Not one line of it said who she was or what she’d done before. The rival’s slides arrived already attached to a name the room recognized. Hers arrived attached to nothing.

Do all of this on every piece going forward, not once as a finished project. One bio page is a start. A steady habit of attaching, matching, and confirming your identity across everything you publish is the actual practice — and it builds on itself, because each new piece corroborates the last one instead of starting the resolution over from zero.

Verify it worked the plain way: ask an assistant a question at the center of your field and see whether it can state, specifically and correctly, what you’ve built. Done looks like a stranger’s tool naming your track record without you in the room. Skipped looks like your best ideas surviving inside someone else’s answer, credited to nobody, or to somebody else.

Where This Breaks By Hand

One page, one byline, one piece — a single afternoon of matching wording and checking the raw source will fix that one piece. Nobody needs help doing it once.

The problem is scale. Every new thing you publish needs the same identity attached, the same sentence matching, word for word, wherever it appears. One inconsistent bio on one guest post fragments you again. One directory that quietly edits your listing, one mention that ages out, one platform that changes how it renders credentials — and the identity you spent months resolving starts drifting apart, usually without you noticing until a competitor’s citation sends you looking for the reason.

Multiply that by every surface where your work might turn up — a guest post here, a mention there, a directory listing somewhere else entirely — and the number of places that can quietly drift out of sync stops being something you can hold in your head. Each one, alone, looks trivial. All of them, unwatched, add up to the exact fragmentation this talk is asking you to close.

Corroboration doesn’t stay put either. Links break. Mentions get buried by redesigns you had no part in. And the systems doing the resolving keep changing how they weigh all of it, on their own schedule, so what confirmed you six months ago is not guaranteed to confirm you now.

None of this announces itself. A directory doesn’t notify you when it drops a field from your listing. A platform doesn’t tell you it changed how it renders your credentials this quarter. The only signal you usually get is the quiet one: a question in your field gets asked, and your name isn’t in the answer anymore, and you have no way of knowing which of a dozen small drifts caused it.

Checking all of that by hand, across everything you’ve ever published, on a schedule tight enough to catch drift before it costs you a citation — that is where the manual version of this breaks. Not the idea. The upkeep.

The people who stay cited aren’t the ones who understood entity consistency better than you did. They’re the ones running something that watches it continuously, so drift gets caught the week it happens, not the week they lose the account and go looking for why.

The Window, Named

The report that lost to a worse one lost for one plain reason: nobody in that room could confirm, in the two seconds they had, who had written it. The system deciding what to surface next week runs the identical check, only faster and colder.

This was never a question of volume, and it was never vanity. Withholding your name from your own good work looks like modesty and functions like erasure. It leaves a machine with nothing to confirm, and someone else’s average work standing in the gap that was supposed to be yours.

There’s a fuller question underneath this one — whether asking to be seen is its own kind of greed — and it deserves more room than a single talk can give it. Today’s answer is narrower, and it’s already actionable: attach yourself to what you’ve built, consistently, everywhere it appears. Pick one piece you published this year and add that one sentence today.

Nobody voted on your behalf to keep that sentence out. You made that call alone, quietly, and called it modesty. The people who needed to find you never got a say in the decision.

Be the source, not the echo. That discipline is what this entity work is building toward — the far edge of the same practice this talk only opens the door to.

The work speaks. But someone has to open the window so the world can hear it.

And once your name is finally attached to your work, one thing remains unresolved: everything else ever said about you, elsewhere, that you didn’t write and can’t take back — read by the same machines, whether you meant it to speak or not.

What have you published this month that still doesn’t say, anywhere, who made it?