X computes a public “About this account” panel for every user: where the account is based, which country’s app store it connects through, whether the platform can confirm its location, how many times it has changed its handle. This study opened that panel for 125,747 amplifiers — the accounts that reply to, repost, and quote the posts driving a conversation — and read those signals back. On each account they sit one tap away. Summed across a post’s crowd, they describe who is carrying the conversation and from where. That summed view lives on the account panel and nowhere a reader in the feed can see it.
Read in aggregate, the origin of the amplifying crowd tracks a subject’s global stakes. A state governor’s crowd is 4.6% based abroad; the President’s is 35.4%; the most internationalized topic in American politics reaches 52.2%. Global salience sets the level: “Israel,” the presidency, and tariffs draw the largest foreign crowds, while topics with mainly domestic stakes — “midterms,” “MAGA,” “NASCAR” — stay close to home.
Where the signals agree, they establish an amplifier’s origin. Where they diverge, the divergence carries information of its own: the accounts the platform places only in a region rather than a country change their handles three times as often as fully concordant domestic accounts, subscribe to Premium at the highest rate of any group, and carry roughly double their headcount in engagement. Each of these facts is computed per account and shown on the account. Summed on the post, they would show a reader the same structure this study assembled by hand.
Each row is one American subject; its bar shows the origin composition of the crowd amplifying it, and the figure at right is the share based in a named foreign country. The tabs redraw the same rows under any single signal or all signals read together; the non-US anchors below mark what a genuinely foreign crowd looks like. Select any row for its full breakdown.
Each US-political subject is placed left to right by how much of the world holds a stake in it — a state race at one end, a geopolitical flashpoint at the other — scored on world-consequence independently of any measured value. Choose an overlay to plot that signal up the vertical axis. Circles are accounts; squares are topics; each is sized by the crowd behind it.
Under the foreign overlay the crowd climbs the continuum: the further a subject reaches into the world’s concerns, the larger the share of its amplifiers based abroad — from 5% at a state race to 52% on the most internationalized topic. The authenticity overlays sort by shape instead of position: account subjects, which carry a repost stream, rise on new and low-follower accounts while topics hold low, and the level moves little along the continuum. Origin follows global stakes; authenticity follows the amplification channel.
Origin composition of each subject’s amplifiers, by country. Allied democracies lead the foreign crowd on every American subject; engagement-farm hub countries form a consistent minority, shown at right as a share of all amplifiers and as a share of the foreign crowd.
The United Kingdom, Canada, Australia, Japan and Germany head the foreign column across American subjects. Hub countries — Nigeria first, then Turkey, the Philippines, Pakistan — run near a fifth of the foreign crowd on the most globally-charged political subjects and rise to two-thirds of it on one entertainment account, where the foreign presence is small but almost entirely hub-based. For scale, the non-US anchors carry foreign shares of 77–91%, drawn overwhelmingly from a single home country each.
The same subjects, split by how an account amplified. Top-engaged replies are written by established, subscribed accounts. The repost stream — the accounts that forward a post — is younger, lower-follower, and more often based abroad. Reposts were sampled for account subjects; keyword topics were sampled by reply only.
On political accounts, 80% of top-engaged repliers carry Premium and the median account is four years old, while the repost stream runs 23% new accounts, 30% under ten followers, and 22% based abroad. Non-political accounts show the same split more sharply: their repost stream is younger still (median two years) at 36% new and 39% under ten followers. The reply stream selects for the established; the repost stream is where new and low-follower accounts concentrate.
Foreign share of the reply stream, by how long after publication each reply arrived. Six subjects on one axis, replies only — the one amplifier type the platform timestamps.
For a minority of amplifiers, the panel names a region rather than a country — “North America,” “Europe,” “West Asia” — or leaves the field blank. This is the least attributable tier in the data, and it behaves distinctly. Among mature (one year or older) US-political amplifiers, region-only accounts have changed their handle three or more times at three times the rate of fully concordant domestic accounts, subscribe to Premium at the highest rate of any group, and carry twice their headcount in engagement. Identity change, geographic non-attribution, and monetization travel together here — each computed by the platform, each shown on the account, none summed where a reader would meet them.
Among the six US-political accounts, amplifiers claim a US location the platform cannot confirm and are under sixty days old — the median is a few weeks. Their behavior is uniform: 99.3% of their amplification is reposting, 97.6% hold fewer than ten followers, and 1.7% subscribe to Premium. They appear in 0.1% of top-engaged replies and in the recent-repost stream at the densities at right. Their share of a subject’s crowd runs highest on the accounts built around a single figure.
At this signal depth, a brand-new account that only forwards and a manufactured amplifier look identical. The facts that would separate them — creation-time clustering, device and network provenance, repost timestamps — are computed by the platform and exposed nowhere. The account-age field alone places this entire population within its first weeks of life.
Keyword subjects (quoted, e.g. “tariffs”) sample the top-engaged replies to the top 20 posts mentioning the term. Account subjects sample replies, most-recent reposts, and quotes on the account’s top 20 posts. The repost stream therefore exists for account subjects; topics are measured on replies alone, and the two-crowd and silent-stream findings are account-scoped for that reason.
These are the platform’s casual, public signals. An operator behind a residential proxy reads as domestic and sits inside the confirmed-domestic tier. Every foreign and region-only figure is a lower bound; the full count is measurable by the platform alone.
63% of accounts created this year carry an unconfirmed location, against a flat ~17% at every other age. The flag primarily marks account age. It is read here as a disclosure state, and the silent-stream population is defined partly by newness, so its unconfirmed status is expected.
Replies were sampled by engagement and reposts by recency. Near-zero-follower and brand-new accounts rarely win top-engaged reply slots, which is why they surface in the repost stream and stay near 1–2% in the timed reply data. The timing chart therefore reports foreign and handle-churn arrival, which the reply stream does contain.
Engagement weighting uses replies and quotes; reposts carry no engagement figure or timestamp in public data. One subject’s foreign engagement share rests on a single viral reply from an Argentine sports account (marked †); one keyword has n=139 (marked ‡).
“Israel” draws the most foreign crowd on the board, and its foreign repliers skew established and verified. At this depth, diaspora and genuine global interest read the same as an operation. Every figure here describes visibility of origin, and each signal is shown separately because the signals disagree, and reading them together is what the disagreements reveal.