# X Open Sources Its Ranking Algorithm > [!question] Lingering Questions > - Is the code on GitHub the actual production ranking logic at any given moment, or a representative-but-lagging snapshot — and how would an outside party verify which? > - What counts as a "label" in the exported JSON — visibility reduction, a moderation strike, a fact-check attachment, something else — and is that definition documented anywhere the average user will find it? > - What are X's actual criteria for merging an external pull request, and has any external PR been merged since the January 2026 disclosure? > [!tip] Article Ideas > - Compare X's PR-acceptance opacity (no stated merge criteria) against The WPOCC's open-source governance transparency goals — a short case-study note on "open code, discretionary merge" as a governance pattern other orgs may cite as precedent. #articleseed > [!link] Cross Reference > - [[3. Resources/Permanent Notes/The Stadium Is the Dominant Open Source Pattern]] — adjacent but distinct: that note tracks contributor concentration, this source's "open code, discretionary merge" model is a related but separate governance-transparency question > - [[3. Resources/MOCs/Open Source Governance MOC]] > - [[3. Resources/MOCs/Commons & Community Governance MOC]] > [!quote] Vocabulary > - **Shadowbanning** — reducing a user's or post's visibility without notifying them, a live accusation against Twitter/X since at least 2018. > - **Under the Hood** — X's new self-service settings page letting eligible users download a JSON export showing whether labels were applied to their account or posts in the past month. ## Bib capsule TechCrunch, "X open sources its ranking algorithm, letting users see if they've been 'shadowbanned,'" August 13, 2026. https://techcrunch.com/2026/08/13/x-open-sources-its-ranking-algorithm-letting-users-see-if-theyve-been-shadowbanned/ ## Author's thesis (Q1+Q2) X released a much larger open-source drop of its "For You" ranking stack (roughly 10-15x its prior January 2026 disclosure), paired with a self-service "Under the Hood" export letting eligible users check whether labels were applied to their account. This is a genuine expansion of algorithmic transparency, but a narrow one: the Grok-based rule-violation predictor stays closed, and the code-level openness runs alongside continued opacity in company-level reporting (no SEC filings since going private, less frequent disclosure of user metrics, revenue, and government takedown requests). ## Key arguments (≤3) 1. **Code-level transparency and corporate transparency are moving in opposite directions.** X expanded what's inspectable about its ranking algorithm while corporate-disclosure dimensions (SEC filings, revenue, takedown-request reporting) have declined since it left public-company status. 2. **The "Under the Hood" export operationalizes "am I shadowbanned" into a checkable artifact.** For the first time, users can check rather than take the company's word for it — though access is pilot-gated to accounts a year or older with 10+ monthly posts. 3. **The most consequential system stays closed.** The Grok-based rule-violation predictor — the system most directly tied to what actually gets suppressed and why — is deliberately excluded, explicitly to deny bad actors a map around moderation, which caps how far the transparency claim can reach. ## Claim atoms ### X excludes Grok moderation predictor from open-source release > [!quote] ID:rw_h1 > a few systems are not included in this release, like those that use Grok to predict whether a post could be violating a rule Where/why: Describing what's carved out of the open-source drop — matters because it's the one system most directly tied to what actually gets suppressed. ### Under the Hood export lets users check for account labels > [!quote] ID:rw_h2 > will let users who have posted ten or more times over the past month download their aggregate stats as a JSON file. The file will show if any labels have been applied to their account or posts over the past calendar month Where/why: The mechanism turning "am I shadowbanned" from a suspicion into a checkable artifact — the article's central news hook. ### X's corporate transparency has declined since going private > [!quote] ID:rw_h3 > No longer required to report to the SEC, X isn't as forthcoming in other areas, like its user metrics, growth, revenue, or government takedown requests, which are now less frequently reported Where/why: The article's own skeptical counterweight to the headline transparency framing — code transparency and corporate transparency moving in opposite directions. ## Permanent candidates - **New**: code-level transparency and corporate disclosure as separable axes (P1) is not yet captured — promoted below as [[3. Resources/Permanent Notes/Algorithmic Transparency and Corporate Disclosure Move on Separate Axes]]. ## Processing checklist - [x] Checked for existing Literature Note — none found - [x] Checked for existing Permanent Notes covering this source's claims — no direct match for the transparency-axes claim; promoted - [x] Connected into [[3. Resources/MOCs/Open Source Governance MOC]] and [[3. Resources/MOCs/Commons & Community Governance MOC]] ## Links - [[3. Resources/MOCs/Open Source Governance MOC]] — open-code/discretionary-merge governance model, algorithmic transparency - [[3. Resources/MOCs/Commons & Community Governance MOC]] — platform accountability and user-facing transparency tooling