The 200 Sources That Decide Who You Are
You do not need to monitor the entire internet. You need to monitor the few hundred sources that actually get retrieved when someone looks you up — and that list is far shorter, far more stable, and far more knowable than anyone selling you monitoring wants to admit.
There is a comforting lie at the center of the reputation industry, and it is that the internet is infinite. Infinite means unknowable. Unknowable means unmanageable. And unmanageable means you should pay someone a retainer forever, because the work never ends and the scope can never be closed.
The lie is useful to vendors and corrosive to everyone else. It is also, as it happens, false.
Retrieval is finite
When someone forms an impression of you — a customer, an investor, a journalist, a hiring committee, a regulator — they do not read the internet. They issue a query. A search, a prompt, a name typed into a due-diligence tool. Something on the other end assembles an answer from a small set of sources it considers worth retrieving.
That set is not infinite. It is not even large. Enumerate the sources that meaningfully contribute to a retrieved answer about any given entity and the count lands in the hundreds, not the millions — and for an individual, often in the dozens. The number in the title of this essay is a round figure chosen to make a point about order of magnitude, not a measurement. The argument does not depend on it being two hundred rather than four.
Everything else — the scraped aggregators, the dead directory listings, the syndicated press release on its fourteenth republication — exists, but it is never retrieved. It occupies storage, not perception. Monitoring it is a way of appearing thorough while accomplishing nothing.
Why the number is smaller than you think
Retrieval systems are aggressively economical. A search engine returns ten results and most people read three. A language model assembles an answer from a handful of high-confidence passages and discards the rest. A due-diligence analyst checks six places and stops.
Each of these systems is also deeply conservative about which sources it trusts. They converge on the same small canon: the entity's own site, the encyclopedia entry, the structured record, two or three news outlets with institutional weight, the dominant review platform for the category, and whichever community forum the model has learned to treat as a proxy for honest opinion.
Convergence is the whole point. It means the surface you actually have to defend is small enough to enumerate on a single page.
You are not fighting the internet. You are fighting a reading list — and reading lists can be audited.
Finding your list
The work is unglamorous. You issue the queries your audiences actually issue — not the flattering ones, the real ones, including the ones with the word scam or lawsuit or alternatives appended. You do it across every retrieval surface that matters, not just the one you are comfortable with. You record what comes back, and you record what got cited to produce it.
Then you do it again next month, because the list moves. Not chaotically — slowly, with a rhythm you can learn. A source that appears in eight of ten answers is load-bearing. A source that appears once is noise. The distinction is measurable, and it is the only distinction that matters.
What emerges is not a dashboard of mentions. It is a map of the specific documents, records, and threads that are doing the work of describing you to the world.
What changes when you know the list
Everything downstream becomes tractable. Priorities stop being a matter of taste. If a single Wikidata property is feeding four different systems and it is wrong, that is not one problem — it is the highest-leverage problem you have, and its position on the list tells you so without anyone needing to argue for it in a meeting.
Budget stops being a hostage to vagueness. You are no longer buying coverage. You are buying the correction of a specific, enumerated set of defects on a specific, enumerated set of sources.
And the work acquires an ending. Not a permanent ending — the list drifts, and drift requires attention. But the panicked, unbounded, spend-until-it-feels-better posture that defines most reputation work simply dissolves once you can see the shape of the thing you are managing.
Kevin Kelly's argument for creators was that you do not need a mass audience, only a small and knowable one. The inverse holds here. You do not need to control the whole internet. You need to know your two hundred — and then you need to be right about them.