X Says You Can Now Check Whether It Is Limiting Your Reach
Every platform has a contingent of users who are certain the algorithm has it in for them. X just gave that contingent something to work with — not an answer, exactly, but a way to gather evidence.
The company expanded what it publishes on GitHub to cover the code that decides how posts rank for visibility in the For You timeline. Alongside it, it's testing a tool that shows you which labels have been stuck on your own account and posts.
The two pieces
The first is an expansion of X's existing public code listing. It now goes into more detail on post display ranking, including the content labels that can affect how far something travels.
The second is the one an ordinary user might care about: a tool at x.com/i/under_the_hood that lets you download the labels or tags applied to your posts over the past month. X is testing it with a randomised set of accounts at least a year old, so most people won't have it yet.
"We want people to be able to answer for themselves whether a platform is limiting their reach, whether the system is fair, and why they see particular content," the company said.
The suggested method says a lot
X VP of Product Keith Coleman spelled out how to actually use the release: drop the published source code into a large language model along with your account data, and ask it what's affecting your posts.
Read that again, because it's a fairly honest admission. The code is public. It is not readable by the people who most want to read it. The company's own advice is to hire an AI to translate.
What a shadowban actually is
Two ideas get mashed together here and it's worth pulling them apart. The folk version — a silent, total blackout on an account — mostly isn't how this works. What does exist is a set of account and post-level restrictions triggered by various parameters, which can quietly shrink your distribution without deleting a thing.
That's exactly why the labels tool might matter more than the code. "This post carried this label" is a fact. "I read the ranking code and concluded I'm being throttled" is a vibe.
Who benefits
Right now, honestly, almost nobody. It takes manual work, the tool is in limited testing, and the results need interpreting.
The exception is creators in X's revenue share program. For them a mysterious drop in reach is a drop in income, and a month of label history is the closest thing to an audit trail they've ever had.
The track record is the problem
X has run this play before. In 2023, a few months after Musk bought Twitter, the company put part of its feed algorithm on GitHub and promised regular updates. What actually shipped was partial, with pieces held back, and it was never updated again. Musk has kept citing it as proof of openness ever since, and last month promised once more to fully open-source the codebase.
So the fair read is: real step, unfinished. The labels tool is new and checkable, which is more than the last attempt offered. Whether the code listing stays current — the exact thing that failed in 2023 — is what to watch over the next few months.
Image: Sanket Mishra, via Pexels
