David Zweig Profile picture
Dec 26, 2022 41 tweets 12 min read Read on X
1. THREAD:

THE TWITTER FILES: HOW TWITTER RIGGED THE COVID DEBATE

– By censoring info that was true but inconvenient to U.S. govt. policy
– By discrediting doctors and other experts who disagreed
– By suppressing ordinary users, including some sharing the CDC’s *own data*
2. So far the Twitter Files have focused on evidence of Twitter’s secret blacklists; how the company functioned as a kind of subsidiary of the FBI; and how execs rewrote the platform’s rules to accommodate their own political desires.
3. What we have yet to cover is Covid. This reporting, for The Free Press, @TheFP, is one piece of that important story.
4. The United States government pressured Twitter and other social media platforms to elevate certain content and suppress other content about Covid-19.
5. Internal files at Twitter that I viewed while on assignment for @TheFP showed that both the Trump and Biden administrations directly pressed Twitter executives to moderate the platform’s pandemic content according to their wishes.
6. At the onset of the pandemic, according to meeting notes, the Trump admin was especially concerned about panic buying. They came looking for “help from the tech companies to combat misinformation” about “runs on grocery stores.” But . . . there were runs on grocery stores. Image
7. It wasn’t just Twitter. The meetings with the Trump White House were also attended by Google, Facebook, Microsoft and others. Image
8. When the Biden admin took over, one of their first meeting requests with Twitter executives was on Covid. The focus was on “anti-vaxxer accounts.” Especially Alex Berenson: Image
9. In the summer of 2021, president Biden said social media companies were “killing people” for allowing vaccine misinformation. Berenson was suspended hours after Biden’s comments, and kicked off the platform the following month.
10. Berenson sued (and then settled with) Twitter. In the legal process Twitter was compelled to release certain internal communications, which showed direct White House pressure on the company to take action on Berenson.

​​alexberenson.substack.com/p/jesse-jackso… Image
11. A December 2022 summary of meetings with the White House by Lauren Culbertson, Twitter’s Head of U.S. Public Policy, adds new evidence of the White House’s pressure campaign, and cements that it repeatedly attempted to directly influence the platform.
12. Culbertson wrote that the Biden team was “very angry” that Twitter had not been more aggressive in deplatforming multiple accounts. They wanted Twitter to do more. Image
13. Twitter executives did not fully capitulate to the Biden team’s wishes. An extensive review of internal communications at the company revealed employees often debating moderation cases in great detail, and with more care than was shown by the government toward free speech.
14. But Twitter did suppress views—many from doctors and scientific experts—that conflicted with the official positions of the White House. As a result, legitimate findings and questions that would have expanded the public debate went missing.
15. There were three serious problems with Twitter’s process:

First, much of the content moderation was conducted by bots, trained on machine learning and AI – impressive in their engineering, yet still too crude for such nuanced work.
16. Second, contractors, in places like the Philippines, also moderated content. They were given decision trees to aid in the process, but tasking non experts to adjudicate tweets on complex topics like myocarditis and mask efficacy data was destined for a significant error rate example template—deactivate...
17 Third, most importantly, the buck stopped with higher level employees at Twitter who chose the inputs for the bots and decision trees, and subjectively decided escalated cases and suspensions. As it is with all people and institutions, there was individual and collective bias
18. With Covid, this bias bent heavily toward establishment dogmas.
19. Inevitably, dissident yet legitimate content was labeled as misinformation, and the accounts of doctors and others were suspended both for tweeting opinions and demonstrably true information.
20. Exhibit A: Dr. Martin Kulldorff, an epidemiologist at Harvard Medical School, tweeted views at odds with US public health authorities and the American left, the political affiliation of nearly the entire staff at Twitter. Image
21. Internal emails show an “intent to action” by a moderator, saying Kulldorff’s tweet violated the company’s Covid-19 misinformation policy and claimed he shared “false information.” Image
22. But Kulldorff’s statement was an expert’s opinion—one which also happened to be in line with vaccine policies in numerous other countries. Yet it was deemed “false information” by Twitter moderators merely because it differed from CDC guidelines.
23. After Twitter took action, Kulldorff’s tweet was slapped with a “Misleading” label and all replies and likes were shut off, throttling the tweet’s ability to be seen and shared by many people, the ostensible core function of the platform: Image
24. In my review of internal files, I found countless instances of tweets labeled as “misleading” or taken down entirely, sometimes triggering account suspensions, simply because they veered from CDC guidance or differed from establishment views.
25. A tweet by @KelleyKga, a self-proclaimed public health fact checker, with 18K followers, was flagged as “Misleading,” and replies and likes disabled, even though it displayed the CDC’s *own data.* Image
26. Internal records showed that a bot had flagged the tweet, and that it received many “tattles” (what the system amusingly called reports from users). That triggered a manual review by a human who– despite the tweet showing actual CDC data–nevertheless labeled it “Misleading”
27. Tellingly, the tweet by @KelleyKga that was labeled “Misleading” was a reply to a tweet that contained actual misinformation.

Covid has never been the leading cause of death from disease in children. Yet that tweet remains on the platform, and without a “misleading” label. Image
28. Whether by humans or algorithms, content that was contrarian but true was still subject to getting flagged or suppressed

This tweet was labeled “Misleading,” even though the owner of this account, @_euzebiusz_, a physician, was referring to the results of a published study Image
29. Andrew Bostom, a Rhode Island physician, was permanently suspended from Twitter after receiving multiple strikes for misinformation. One of his strikes was for a tweet referring to the results from a peer reviewed study on mRNA vaccines. Image
30. A review of Twitter log files revealed that an internal audit, conducted after Bostom’s attorney contacted Twitter, found that only 1 of Bostom’s 5 violations were valid. Image
31. The one Bostom tweet found to still be in violation cited data that was legitimate but inconvenient to the public health establishment’s narrative about the risks of flu versus Covid in children. Image
32. That this tweet was not only flagged by a bot, but its violation manually affirmed by a staff member is telling of both the algorithmic and human bias at play. Bostom’s account was suspended for months and was finally restored on Christmas Day.
33. Another example of human bias run amok was the reaction to this tweet by Trump. Many Trump tweets led to extensive internal debates, and this one was no different. Image
34. In a surreal exchange, Jim Baker, at the time Twitter’s Deputy General Counsel, asks why telling people to not be afraid wasn’t a violation of Twitter’s Covid-19 misinformation policy. Image
35. Yoel Roth, Twitter’s former head of Trust & Safety, had to explain that optimism wasn’t misinformation. Image
36. Remember @KelleyKga with the CDC data tweet? Twitter’s response to her is clarifying: “we will prioritize review and labeling of content that could lead to increased exposure or transmission.” Image
37. Twitter made a decision, via the political leanings of senior staff, and govt pressure, that the public health authorities’ approach to the pandemic – prioritizing mitigation over other concerns – was “The Science” . . .
38. Information that challenged that view, such as showing harms of vaccines, or that could be perceived as downplaying the risks of Covid, especially to children, was subject to moderation, and even suppression. No matter whether such views were correct or adopted abroad.
39. What might this pandemic and its aftermath have looked like if there had been a more open debate on Twitter and other social media platforms—not to mention the mainstream press—about the origins of Covid, about lockdowns, about the true risks of Covid in kids, and much more?
40. Thanks to @ShellenbergerMD, @lwoodhouse, @lhfang and the team @TheFP for their help reporting this story.
41. An expanded version of this thread is available now @TheFP! thefp.com/p/how-twitter-…

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with David Zweig

David Zweig Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @davidzweig

Jan 4, 2023
Merely citing hospitalizations, see original tweet, can be misleading

In the last 8 months severe Covid has decoupled from Covid hospital numbers. The portion of Covid admins in ICU are *half* of what they were earlier in pandemic

And huge difference for >70 vs rest of pop

/1 Image
We've known for a long time that a significant portion of Covid hospitalizations are incidental. I reported on this twice. First, for pediatric numbers, in May 2021

/2

nymag.com/intelligencer/…
And again for adults in Sept 2021

/3

theatlantic.com/health/archive…
Read 5 tweets
Nov 29, 2022
THREAD

UK Report on masks in schools

*75% of teachers said masks made communicating harder for special needs kids
*Teachers said masks made teaching languages harder
*Masks muffled sound, making communication hard for teachers and kids hard of hearing
/1

assets.publishing.service.gov.uk/government/upl…
*Masks were an issue with kids who needed to lip read (clear masks did not resolve issue)
*Teachers struggled to pick up non verbal cues from masked autistic children
*Masks enabled quieter pupils to not participate as much

/2
*1 in 5 pupils said wearing a mask made them feel anxious, with special needs kids overrepresented in this group
*Teachers said masks made it hard to see which kids were struggling
*Students noticed acne and skin issues from mask wearing

/3
Read 5 tweets
Oct 25, 2022
The NAEP scores and NYT article are misleading a lot of people to conclude remote school didn't correlate with lower scores. That's bc state data are too crude

Multiple analyses comparing districts show an overt correlation btwn time out of school and learning loss

Thread: /1
This study by @vkoganpolisci in Ohio found "Districts with fully remote instruction experienced test scores declines up to three times greater than districts that had in-person instruction"

glenn.osu.edu/how-covid-19-p…
This paper from
@ProfEmilyOster et al (see "% in person")
emilyoster.net/wp-content/upl…
Read 7 tweets
Sep 5, 2022
Short 🧵

This thread purporting that school closures/remote learning was not correlated with learning loss unsurprisingly has gotten a lot of traction. Yet it likely suffers from some major flaws and limitations, calling its conclusion into question

/1
Primarily, state-level data, which the author relies on, is overly crude. At the local and district level, numerous, more rigorous analyses have found an overt correlation between time out of school and learning loss
Eg A: This study in Ohio found "Districts with fully remote instruction experienced test scores declines up to three times greater than districts that had in-person instruction"

glenn.osu.edu/how-covid-19-p…
Read 6 tweets
Aug 10, 2021
THREAD

I corresponded with the authors of the Duke study behind this opinion piece weeks ago... nytimes.com/2021/08/10/opi…
A topline finding of the Duke/ABC Science Collaborative report, and what they claim here, is that masks in schools help lower transmission. Except all their schools were under a mask mandate. I asked the authors how they could make a claim on masks when there was no control ...
In order to claim an effect you must compare one group with an intervention to a different group *without* the intervention, which they didn't do. Instead, the authors replied with the Israeli study, saying that showed masks in schools work. Except...
Read 9 tweets
Feb 18, 2021
💥🔥🧨 I am beyond excited to announce that I will be writing AN ABUNDANCE OF CAUTION, a book for @mitpress, about the schools debacle in America during the pandemic. The book, in essence, will be a case study of how and why bad decisions are made

THREAD
1/12
I first covered the evidence on children and schools on May 11, 2020, which, contra claims by some politicians and journalists, was highly suggestive even then (after all, it was convincing enough for Europe)

2/12
wired.com/story/the-case…
It is hard to overestimate how much of an anomaly this article was at the time

3/12

Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(