Positioning in a World of Infinite Answers

Ask any AI assistant the central question of your field right now. You’ll get a fair, accurate, free answer in about four seconds — so what exactly are you for?

The Cost of Staying Fair

Sit with that question before you reach for a comfortable answer, because the comfortable answer is almost always wrong. You are not for holding information anymore. Nobody is. A machine can now summarize the standard approaches in your field, cite the consensus, and hedge it fairly, faster than you can open a document to start writing the same thing.

The actual cost of ignoring this is plain. If everything you publish is a balanced overview — the five approaches, the pros and cons, the fair-minded roundup — you are producing exactly what the machine already produces for free. It will fold your careful summary into the same average it builds from a dozen other careful summaries, and it will have no particular reason to say your name while doing it. You won’t get a rejection. You’ll never get named. Every month you spend being scrupulously fair is a month somebody else’s sharper, riskier claim gets cited in the place that should have been yours.

That is the stake. Not fewer visitors. Fewer reasons for anyone — human or machine — to reach for you specifically. A wire service reporting the same vote as every other wire service adds nothing a reader couldn’t get elsewhere; it disappears behind the facts by design. That was once respectable, paid work. Now it is exactly the shape of work a machine performs for free, in seconds, without needing your byline at all.

Why the Room Names Some Voices and Averages the Rest

There’s a process behind every answer a machine gives a stranger — pulling from whatever sources it can find, weighing how much they agree with each other, and deciding what to say. The full course walks that process end to end, from how it first gathers competing sources to how it eventually seats one voice for good. Tonight you only need one piece of it: how that process treats agreement versus disagreement.

When ten sources say roughly the same fair, reasonable thing about a topic, the machine has no reason to prefer any one of them. It corroborates the consensus and reports it back as consensus, unattributed, because attributing an average to any single name would be inaccurate. Sameness gets synthesized. It does not get cited.

When one source says something specific, consistent, and different from the average — a claim it has made the same way, in its own name, across multiple places the machine can independently find — something else happens. The machine can no longer fold that claim into the consensus without distorting it. The honest move, and the one these systems are increasingly built to make, is to attribute it: to say who holds this view, because the view itself cannot be produced by averaging everyone else.

This is the mechanism worth naming plainly: attribution requires distinction, and distinction requires a position the average cannot generate. A topic summarized fairly is, by definition, reproducible by anything else that summarizes it fairly. A specific claim, repeated consistently and traceable back to one name across independent pages, is not reproducible that way — it can only be repeated, and repetition is what citation looks like.

You can see the same pattern in any field mature enough to have named schools of thought. Practitioners in a mature field don’t split into “the ones who know the material” and “the ones who don’t” — the material is shared property, available to anyone who looks. They split into camps, each built around a distinct, defended interpretation of the same shared facts, specific enough that other people eventually need a name for it, even to argue against it. A named school of thought is a position that got repeated consistently enough, in one name, for long enough, that it stopped being an opinion and became a reference point.

Topics are taken. Points of view are wide open. That’s the whole talk, said once, early, so you can watch the case for it build.

Building the Position the Room Can’t Average Out

Start by writing the position itself, in one sentence, before touching anything else. Not “here are five approaches” — a specific, falsifiable claim you’re willing to defend: most people in your field believe X, and you believe that’s wrong, because Y. If you can’t finish that sentence yet, nothing downstream will matter, because there is no distinct claim yet for anything to attribute.

Once you have the sentence, publish it somewhere durable — one clearly-marked page that states it plainly, near the top, in ordinary language, not buried under three paragraphs of scene-setting. This becomes your canonical statement of the position: the page you’d point to if someone asked what you believe. Make sure it survives a script-free view of the page — the same render check that matters for any page you want a machine to read at all. A bold position sitting inside a script-only element is invisible to a fetcher no matter how boldly you meant it.

Then repeat the position, consistently, everywhere your name appears — your bio, your about page, guest posts, interviews, any place a machine might independently find you. Consistency here isn’t stylistic pickiness. It’s the difference between a machine recognizing one claim showing up in several places and a machine seeing several vaguely similar claims it has no reason to connect. If your bio says one thing and your guest post says something close but not identical, you’ve made corroboration harder, not easier, because the systems doing this work are matching claims, not vibes.

Structured markup helps here in a small, specific way: author and entity markup that consistently identifies who you are across the pages carrying your position gives machines an easier path to recognizing that the claim on page one and the claim on page twelve belong to the same person. It doesn’t create the position. It only keeps your name from fragmenting into several less-recognizable versions of itself across the web.

Corroboration matters more than volume here. One canonical page stating your position, echoed accurately in three independent places other people control — a podcast transcript, an interview, a piece someone else wrote citing you — outweighs twenty restatements of it on properties you own alone. Independent agreement is what tells a synthesis system this claim belongs to a real, external, checkable entity, not only a claim you keep repeating to yourself.

Next, check whether the position landed. Pose your field’s central question to a machine and read past the headline into the full paragraph. Notice whether your name shows up attached to a specific claim, or whether you’ve dissolved into “some experts believe” like everyone else who stayed fair. That gap tells you plainly whether your position has cleared the bar or is still reading as one more balanced summary.

Keep the position current. A claim that goes untouched for years starts to look abandoned rather than defended, and freshness signals matter to systems deciding what’s still live enough to cite. Restate it, defend it against new evidence, update the page when the terrain shifts — not by softening it, but by showing the position survived contact with something new.

What breaks when any of this is skipped is predictable. Skip the plain, findable canonical statement, and there’s no single page a machine can point to when it wants to attribute the claim. Skip consistency, and your one position reads as three vague, unconnected ones. Skip verification, and you’ll never know whether the room is naming you or quietly averaging you into the fog. Done looks like this: ask the honest question, and your name arrives attached to a specific claim, not listed as one interchangeable source among several.

Where Doing This by Hand Runs Out

None of this needs anyone’s blessing or a budget. Write the sentence tonight, put it on a page, and go check whether one assistant repeats it back with your name on it — that’s an evening, and it’s entirely yours to do.

Here’s where the arithmetic turns against you. One position, checked once, is an evening’s work. But your position doesn’t sit on one page — it sits on every bio, every guest post, every interview transcript, every place your name gets picked up and occasionally re-worded by someone else entirely. New content about your field is published constantly, which means the average your position is competing against is never the same average twice. A phrase you tightened for clarity last year can drift out of sync with an older version still live on a syndicated post you forgot about, and nothing announces that the drift happened.

Answer engines update. Models retrain. A position that got attributed to you in the spring can quietly stop being attributed by autumn, for reasons that have nothing to do with whether you still believe it. The only way to know is to keep asking, across a growing list of assistants, on a schedule you don’t forget — and to keep checking that every page carrying your name still says the position the same way.

That’s the actual wall. Not the difficulty of holding a view — that part is only spine. The difficulty is that verifying, across everything you publish, that the position still reads as one consistent, current, attributable claim is a task with no natural end date. Meanwhile the field doesn’t wait for you to catch up. Someone else in your space is publishing their sharper, riskier claim this week, consistently, everywhere, while you’re still deciding whether to check your own. The people winning this right now aren’t holding sharper opinions than everyone else. They’re the ones who stopped treating verification as a one-time check and built a way to keep watching after they stopped looking themselves.

Sharpened, Not Shattered

So: when every question has a free, instant, decent answer, what exactly are you for?

Not for holding information — that stopped being scarce the day machines learned to summarize it for free. You’re for holding a position: specific, tested, consistently defended, the kind of claim an average of everyone else’s opinions structurally cannot produce, because producing it requires being willing to be wrong, in public, and keep standing there anyway.

Topics are taken. Points of view are wide open. Say it once more and notice what it asks of you now that it didn’t at the start of this talk: not choosing a topic to know, but choosing, on the record, to mean something specific about it.

Carry one question into this week: if a stranger asked a machine tonight not what your field believes, but what you believe — is there a real answer, specific enough to disagree with?

There’s a harder question waiting past this one. If a position is what stays scarce once information is free, what happens once even your position can be copied, paraphrased, and repeated by someone who never earned it? The full course opens that door next.