Facebook says it accidentally removed thousands of posts around the Jan. 6 riots from CrowdTangle, blaming a now-fixed error that has existed since at least May
The social media company said it was a technical error, and has been fixed, but tens of thousands of posts are still missing. Tweets: @markscott82 , @natematias , @markscott82 , @jason_kint , @profcarroll , @giladedelman , @sayhar , @evelyndouek , @evelyndouek , @individual1isa1 , @jason_kint , @jason_kint , and @politicoeurope Tweets: Mark Scott / @markscott82 : @Facebook The only option available is to accept this and allow FB to collect data on my activty on other apps/websites. I'm no lawyer. But that's not consent. J. Nathan Matias / @natematias : It's common for research based on platform data to wrestle with problems of missing data, often for very boring reasons. How much should we worry about this missing data problems? A thread. https://twitter.com/... Mark Scott / @markscott82 : Oh this is fun. Tried to sign into @Facebook's Crowdtangle. Get pinged with multiple messages to accept cookies. Turn them all down (except for “essential” ones). Now have this box, and if I don't click allow, I can't sign into FB. Excellent. Excellent. https://twitter.com/... Jason Kint / @jason_kint : wow, this is insane. it's also where I remind those new to Facebook's rodeo that in March/April 2017, while under major scrutiny from an election, Facebook purged millions of active fake accounts that were manipulating news pages then they buried the details. Trust no one there. https://twitter.com/... https://twitter.com/... David Carroll / @profcarroll : The problem with “free (not actually free)” software like Facebook is that their negligence or malign cover-ups can be designated as “bugs” even if it's total bullshit. How are we to know? https://twitter.com/... Gilad Edelman / @giladedelman : @deaneckles Yeah. But even FB describes this as a new level: “we're putting more emphasis on new signals such as how likely people are to provide us with negative feedback on posts about political topics and current events when we rank those types of posts in their News Feed.” Sahar Massachi / @sayhar : I trust @nicole_bonoff on a lot. So I provisionally have to agree with her. I have a different way of seeing this, though: this change should obviously be seen as a bandaid that *probably* won't work, and definitely doesn't solve root problems. https://twitter.com/... Evelyn Douek / @evelyndouek : Every transparency report and tool from every platform could be as accurate as Enron's and we would be none the wiser https://twitter.com/... Evelyn Douek / @evelyndouek : Independent auditing is low hanging fruit. There's self-regulatory moves towards this (driven by advertisers funnily enough) but it needs regulatory backing Lib Lady Army Veteran😷🌊🇺 🇸: @RVAwonk @MrJonCryer So Facebook is deleting content that the @January6thCmte specifically asked for? Isn't that umm, illegal? Jason Kint / @jason_kint : ps for the press, I would humbly recommend avoiding any off record explanations from Facebook. Ask them how many posts are missing and a detailed accounting of them. They already lost the right to not be transparent about them. And don't waste your time with Andy Stone. Jason Kint / @jason_kint : It's also where I remind anyone involved with investigating January 6th, whether in Congress or the press or academia, you cannot trust the company. I would have independent auditors go in, depose those with access and verify everything. Should have done this in 2017. @politicoeurope : Facebook failed to disclose the disappearance of scores of posts from the days before and after the Capitol Hill riots. The revelation comes at a difficult time for the social media network and its efforts to promote transparency. https://www.politico.eu/...
Context & Ripple Effects
CrowdTangle is the main window outside researchers have into what actually spread on Facebook, so an unannounced deletion of Jan. 6-era posts is not just a bug story — it removes primary evidence from exactly the period Facebook's own reviews keep flagging. The company had already conceded internally that it failed to act on Stop the Steal groups before the Capitol riot, and reporting on shelved polarizing-effect research showed how findings that embarrass the platform tend not to surface on their own.
That is why the reaction in the tweet stream runs through disinformation researchers like Evelyn Douek and J. Nathan Matias: when the only archival tool is owned by the company being studied, a 'technical error' persisting silently since May reads as a governance problem, not an engineering one. It lands alongside [[a:970794|internal findings that Eastern European troll farms reached 140M US users a month before the 2020 election]] — more evidence the record needs independent verification.
First-order effects
- Disinformation researchers and journalists lose tens of thousands of still-missing Jan. 6 posts from CrowdTangle, weakening any audit of what Facebook circulated during the riot window.
Second-order effects
- Each self-reported failure hands Congress fresh justification to mandate independent researcher access to platform data, rather than relying on tools like CrowdTangle that Facebook controls and can break.
Third-order effects
- If platform-owned transparency tooling keeps failing this way, the durable outcome is legislated, interface-neutral data access for accredited researchers — with Facebook's own record, including its earlier overhaul of data governance after firing 52 staffers for misusing user data, cited as proof self-policing is insufficient.
The trend: Platform-supplied transparency infrastructure is proving too fragile and conflicted to sustain independent oversight, pushing regulators toward mandatory researcher data access.