How I Almost Went to the Clink Over One Letter

Two years ago I nearly sank the Twitter Files project with a stumbling appearance with Mehdi Hasan, but the real mistake was going on the show at all

Two years ago I made a catastrophic decision to give an interview about the Twitter Files with MSNBC’s Mehdi Hasan. I made a worse decision to cop on-air to what at the time I thought was at most a minor mistake, interpreting a Slack passage reading “According to CIS (escalated via EIP)” as “According to CIS[A] (escalated via EIP).”

The alleged error here was substituting the Cybersecurity and Infrastructure Security Agency, or the Homeland Security body CISA, for the Center for Internet Security, or CIS. As Mehdi put it, I’d “deliberately and under oath” made communications from a “nonprofit” look like recommendations from an “intel agency.” This episode, incredibly, led to a letter from Congress (not even sent to me, oddly) threatening a prison term of “up to five years” for “providing false information” under oath:

I took a bigger beating for a line about the Stanford-run Election Integrity Partnership: “According to the EIP’s own data, it succeeded in getting nearly 22 million tweets labeled in the runup to the 2020 vote.” Stanford issued a spirited defense of the group’s efforts, writing, “No, the EIP did not censor any tweets or label any tweets as ‘misinformation.’” It added that platforms were responsible for “their own content moderation decisions,” and that the EIP “did not make recommendations to the platforms about what actions they should take.”

It later came out (after a long legal fight) that the EIP did send concrete recommendations to platforms in big numbers, saying things like:

• “We recommend Twitter remove the tweet as it is a fairly clear violation”;

• “Hi Facebook, Reddit, and Twitter . . . we recommend it be removed from your platforms”;

• “Recommend disabling the account at link”;

• “We recommend you label or reduce the discoverability of the post.”

There were many more of this type, underscoring a basic problem of the Twitter Files. With the EIP, there was significant public data about its research endeavors, but the flip side was a real-time reporting program thrown together between the platforms and the Department of Homeland Security in 2020 for the purpose of policing tweets in time for that year’s election. We knew it was important because email after email said things like “CISA received a grant to build a web portal for state and local election officials to report incidents of election-related misinformation,” and “DHS want to establish a centralized portal for reporting disinformation.” Once operational, complaints about topics came in via routes like the EIP, after which Twitter would issue quasi-automated content moderation responses in big quantities. But how big?

Yesterday Mike Benz tweeted a new video showing an EIP partner talking about how the group “identified 20+ million tweets related to ~400 false, misleading, or unsubstantiated claims” that “functioned to sow doubt in the election.” The video doesn’t change much, echoing things written in the EIP’s report.

I let this alone for ages because it’s a weedsy matter, because Lee Fang already addressed many of the main issues, and because players like Hasan refused to retract their own errors, even after I was threatened with jail over them. But I want to explain part of this bizarre story.

Ben Kawaller yesterday wrote about the phenomenon of activists refusing to engage with potentially hostile media. Once, I believed journalists should never do the same. I thought reporters should always engage, even if you know you’re going to take a beating. Now, I’m not sure: