Chart Integrity Note — Parasocial Streaming 100 & Surge 50 — Edition 4, first correction (28 July 2026)
Effect on Surge eligibility
Published by Borealis, the data and measurement layer behind the Influence Index charts.
Summary
Edition 4 was published from an incomplete field. A fault in how we read our own measurement data limited scoring to 1,000 creators out of roughly 1,288 active on the roster, so around 275 creators were never scored at all — not ranked low, not filtered out, simply absent from the calculation.
One of them belonged at number three.
The edition has been regenerated over the full roster and republished as version 2. The original version remains permanently on record.
This is a different kind of correction from the one we published on July 6. That one was about a channel that did not belong on the chart. This one is about creators who did belong and were missing — an error in our measurement, not in anyone’s conduct.
What went wrong
Our scoring engine asks the database for every active creator’s weekly totals. The database layer caps any single response at 1,000 rows and — this is the part that made it dangerous — does not report that it truncated. A response of exactly 1,000 rows looks identical to a complete one.
Our roster passed 1,000 creators during the period covered by this edition. From that point the scoring engine was reading a partial field while reporting success. Which creators fell outside the 1,000 was arbitrary: the query had no defined ordering, so it was not the smallest or the newest, just whichever the database returned last.
Because ranking is relative — every creator’s position depends on who else is being measured — a missing creator does not only lose their own place. It shifts everyone below where they would have been.
What changed
Comparing the corrected edition against what was published:
| Streaming 100 | Surge 50 | |
|---|---|---|
| Creators entering the chart | 29 | 28 |
| Creators leaving the chart | 29 | 28 |
| Positions retained | 71 | 22 |
The top of the chart changed materially:
| As published (v1) | Corrected (v2) | |
|---|---|---|
| 1 | Ibai | Ibai |
| 2 | TheBurntPeanut | Jynxzi |
| 3 | BLASTPremier | Suburbbaby |
| 4 | JasonTheWeen | CaseOh |
| 5 | 布団ちゃんと申します | fps_shaka |
Suburbbaby — 3.3 million hours watched over the tracking week — did not appear on the published edition at any position.
The Surge 50 excludes creators who have ever peaked inside the Streaming 100’s top 25. That is a permanent bar, so a change to the top 25 changes who is eligible going forward. Measured across the full corrected edition history:
Eight creators are eligible for the Surge 50 again — their only top-25 placement was one the corrected edition does not reproduce. Two creators are newly graduated out of Surge eligibility on a first-ever top-25 placement.
These are the true figures, computed from each creator’s best position across every active edition. A naive comparison of only this edition’s top 25 would suggest twelve in each direction; most of those creators had already graduated in an earlier edition, or keep an earlier top-25 placement that still stands.
Both sets are derivable from the retained ledger, which holds the superseded and corrected versions of every edition. We are not listing the eight by name: the change is a mechanical consequence of correcting who was measured, and naming them would present a data correction as a judgment about them.
What we did
The correction was made the same way as any other — through the engine, not by hand:
- Regenerated, not edited. The full edition was recomputed by the same scoring engine that produces every edition, over the complete roster. Every score and position re-derived from the underlying measurements. No position was moved manually.
- Nothing was overwritten. The chart ledger is append-only. The original edition and the corrected edition both remain permanently on record; the published pointer now references the corrected version. Any published number, before or after, can be reconstructed.
- The methodology did not change. This edition was published under the rules in force at the time, and it remains stamped with them. We corrected who was measured, not how.
- The scoring path now proves its own completeness. The reads behind ranking page through the full result and check what they received against an independent count of what should have been there. A scoring run that cannot show it saw the whole field stops rather than publishing.
We are describing the path that produced this edition, not claiming the underlying fault is eliminated everywhere. Silent truncation has surfaced repeatedly across this system in different reads, and this edition is not the first time. Remaining reads of that shape are inventoried and enforced against by an automated check that fails the build if a new one appears; they are being converted in order of what they can affect. We would rather state that plainly than claim a clean sweep.
Ledger reference
The reason recorded against the published-version pointer, verbatim, so this note and the database can be checked against each other. Recorded for both streaming-100 and streaming-surge-50 at edition_date = 2026-07-27, active_version = 2, updated_by = integrity-supersede, updated_at = 2026-07-28T20:25:57Z:
Scored population was truncated at the PostgREST 1000-row cap by an unpaged read of compute_creator_period_metrics, omitting roughly 275 of the ~1288 active creators from scoring entirely – including creators ranking inside the top 10. Regenerated on the paged, coverage-gated rollup path over the full roster. methodology_version unchanged at 1: this corrects the population, not the eligibility rules edition 4 was published under.
Standing commitments
- Auditability. Every published number is reconstructible from retained data, including the superseded version of any corrected edition.
- Corrections are versioned, never silent. If a published edition is corrected, the prior version is retained and the correction is recorded with a public note — like this one.
- We publish our own errors. A measurement fault that changed a published chart is disclosed on the same terms as an eligibility decision. The credibility of a chart rests on what it admits as much as on what it excludes.
Questions or disputes about a position or an eligibility decision can be directed to the Borealis integrity desk. This note accompanies the corrected Edition 4 of the Parasocial Streaming 100 and Parasocial Streaming Surge 50, republished July 28, 2026.
Amendment — 30 July 2026
The standing commitment above reads: “Every published number is reconstructible from retained data, including the superseded version of any corrected edition.”
The second half is correct — every superseded version is retained in full, and that is how this edition’s correction is auditable. The word “reconstructible” claimed more than we could deliver.
A creator’s score depends on the field they were measured against: every other eligible creator observed that week. That field was read live from a roster that changes daily and was not captured at publication, so recomputing this edition today does not reproduce it exactly. The arithmetic is unchanged; the set it ran over is not. This is the same class of dependency the note above describes — the comparison set determines the score — which is why the distinction matters here in particular.
We asserted reconstructibility before the mechanism that guarantees it existed. That mechanism exists from the edition of 3 August 2026, which freezes the full comparison set alongside the ranking. Editions from that date forward can be reconstructed exactly as published. This edition cannot.
What is retained for this edition, and permanently:
- Every position and score exactly as published, in every version.
- Every superseded version, in full, with the date each was published.
- The correction described in this note.
A second correction to this edition was published on 30 July 2026, removing a duplicate record that had seated one creator at two positions. It is described in its own note.
Where to send a dispute or a question about this note: the Parasocial charts desk, at corrections@parasocialmagazine.com. What to send, and what happens to it, is set out on the methodology page.
The note above refers to “the Borealis integrity desk” and gives no way to reach it. That is the same desk under its former name — Borealis is the internal measurement layer behind the charts, and naming it to a reader explained nothing. Reader-facing copy now says the Parasocial charts desk.
The byline at the top of this note is wrong on both counts — it names Borealis, and it attributes these charts to the Influence Index, a separate property, when the note concerns the Parasocial Streaming 100 and the Parasocial Streaming Surge 50.
This amendment corrects the note above. The original text is left intact rather than edited, because a note whose subject is publishing our own errors should not correct itself by deleting one.
