The X Factor and Algorithmic Suppression

After Elon Musk completed the purchase of Twitter on October 26th, 2022, he walked into headquarters carrying a sink and a lot of hope from users with conservative leanings. The reason for the hope was that censorship and throttling were deeply rooted in the organization and, by extension, within the algorithm. This was despite denials by Twitter 1.0 senior staff that this was the case. Elon made it clear to everyone on day one that change was coming. By all appearances over the past year, that seemed to be somewhat true.

However, lurking issues within the algorithm are still affecting users today. With complaints of reach to followers, the reach of replies, and general account throttling, among other issues, someone needed to look under the hood. At the time of the purchase, though, that was not possible. Elon, however, in an act of transparency, made parts of the algorithm open source for review via GitHub on March 31st of this year.

With that release, and what this article will cover, is a highly complicated topic for some as the totality of it all is difficult to wrap your head around. But hopefully, we will be able to break this down for you so that you better understand how Twitter, now X, was and still is affecting accounts on the platform.

This is where The Parzival steps in. With a degree in engineering and working knowledge of coding in multiple languages including calculus-based integration into algorithms, he began digging. He started this venture because he noticed that his account and others were not getting the traction they deserved.

His deep dive of the algorithm showed the gears and inner workings of what it does and anyone with technical savvy is able to follow the breadcrumbs he laid out in his pinned thread. What becomes evident is, there are still vestiges of Twitter 1.0 running roughshod over conservative accounts, and it didn’t take him long to discover the first censorship mechanisms. What he found was shocking.

The first discovery was a label called "Tombstoning." Its purpose was hiding or removing tweets based upon a set of rules. Those rules were defined by a "VisibilityResult" object containing information about the tweet, including language and country. If the result was to apply the Tombstone, there would be an epitaph, essentially giving a reason why it killed the visibility. From the user's perspective, the tweet would appear as "Post is unavailable."

But as The Parzival delved through the layers, he eventually uncovered another section labeled "GovernmentRequested." This label was applied directly to topics and or accounts at the behest of government actors with regard to so-called misinformation. This was most likely in response to COVID-19 or election topics they wanted to suppress or outright remove. One thing to note, however, is that it does not mention which government. So technically, these requests could come from any government entity, such as US agencies or agencies in the EU.

It should be noted here that these labels can be automatically, via the algorithm, or manually applied by the Trust and Safety team members to limit reach of accounts or even topics. Below is an example of applied labels to the Libs of TikTok account. One of the hidden labels that seemed more directed and nefarious, however, was if there was a specific user that Twitter didn’t like. This label was called “persona non grata”. Which furthered the limitation of the account it was applied to.

As Elon's team works on these issues, it appears that some sections are planned to be deprecated at some point. These labels included "MisinfoCovid19", "MsnfoUsElection," and the aforementioned "PersonaNonGrata." So, some deliberately marked items are slated for removal to help the users of the X platform. When those changes will be made is unknown.

The next thing on the list to discuss is that Twitter 1.0 was applying a reputation score on its users. This scoring system was based on follow/follower ratios, device usage, safety status (restricted, verified), and account age. Essentially, this would place a user on a scale from 15 to 95 and either boost or reduce a user's reach. You could equate this to a credit score. The higher your score, the better.

The follow/ follower ratio itself was worked out by The Parzival in collaboration with justredpillme. As you can see below, the follower-to-following ratio could be good or bad for your account.

The last thing to highlight in the algorithm is clustering. This part of the algorithm grouped users based on similar topics of interest and following relationships between accounts. Clustering essentially divvied up the user base into categories, allowing Twitter 1.0 to suppress an entire user group or groups at a time. It's where Twitter 1.0 bias shined brightly despite the repeated denials that conservatives were not being targeted. This method was far more effective at censorship due to its broad nature of suppressing groups of accounts and specific controversial topics.

As The Parzival said in his latest thread, “At this point it should be clear that the Twitter algo was designed for 1 purpose and 1 purpose alone, to silence conservatives and control the information narrative on social media.”. That’s the key takeaway here.

Interestingly, as X was making changes to the algorithm, The Parzival was able to denote them as they happened. But he mentions any further alterations stopped as of July 13th. He surmises they likely gave up and decided it was time for a ground-up rewrite. This is potentially promising news considering the recent statement by Elon on November 10th saying a new portion of algorithm was coming and that it would boost smaller accounts outside of your followers.

Tying this all together with the release of the Twitter Files, it wasn’t a far step for Twitter to censor its users at the request of government agencies or even non-government organizations (NGOs), as some of these algorithmic markers were already in place. The recent revelation that Jim Jordan posted in a thread, Election Integrity Partners (EIP) worked in concert with social media platforms such as X, Facebook and others to censor Americans specifically, even before the 2020 election. Some of these censored posts included true information, jokes and opinions.

According to one EIP member, EIP was created "at the request of the Cybersecurity and Infrastructure Security Agency" or CISA. EIP "stakeholders" and government agencies would create misinformation reports, analyze them, find similar content across different platforms, and then submit these reports to BigTech for action. Twitter would take these reports and remove posts, remove users, or apply labels to accounts to reduce reach.

Among accounts affected by government, non-government, and BigTech censorship were President Donald Trump, sitting senators, congressional members, governors, Tom Fitton, Charlie Kirk, and UncoverDC's chief editor, Tracy Beanz. These labels and censorship actions were also applied to organizations like The Babylon Bee, NEWSMAX, and Project Veritas.

One thing is for certain, as more information is revealed on the X algorithm, key people are taking an interest in it. Recently, Representative Anna Paulina Luna asked about the information that The Parzival posted as it could potentially be very valuable to the Weaponization Committee in holding people accountable. Especially the “GovernmentRequested” section and how censorship was being applied at Twitter. Beyond that, one can only imagine that this is still occurring on other social media platforms as well.

The Parzival's X account is worth following if you are interested in seeing what goes on under the hood in relation to the X algorithm. This article could not have been possible without his threads, extensive knowledge, and assistance.

One last thing I will mention is this: we still have the hope Mr. Musk brought into Twitter with that sink as we believe X has the potential to become THE free speech platform. But as we patiently wait for resolution, our frustration is growing. Some X users are at the point where they don't think the $8 is worth it anymore to be verified and turned in their check mark. Hopefully, things will change, and that change can't come soon enough.

David Setliff is a writer who "delivers a worldview from the rural perspective." He is the author of The Flyover Chronicles on Substack. You can also follow Dave on X.

Get the latest news delivered daily!

We will send you breaking news right to your inbox

© 2024