Reporting about the reporting

Hamilton 68, hosted by the Alliance for Securing Democracy, monitored a network of more than 600 accounts during 2017–2018. Taibbi’s January 27, 2023 report published internal Twitter messages sharply questioning the sample. Twitter’s analysis found many ordinary Western users and challenged the inference that their conversations measured a Russian influence operation. That exposed a serious gap between a dashboard’s apparent authority and what Twitter’s own analysts thought the data supported. [1]

ASD replied that the dashboard tracked networks amplifying Russian narratives, including unwitting participants, rather than a roster of people it had verified as Russian agents or bots. Earlier public guidance had indeed warned that the accounts included humans and automation, inside and outside Russia. ASD also acknowledged media misuse of its data and said the original dashboard was retired in December 2018; Hamilton 2.0 shifted to overt, attributable accounts. [2][3]

Two judgments can coexist: Russian origin or control should not be inferred from a hashtag’s popularity within a selected network; yet an American account can circulate state propaganda without being Russian or automated. Taibbi’s sweeping fraud characterization and ASD’s methodological defense should each face the actual sample, selection rules, and specific news report at issue. Neither establishes that all foreign interference was invented. [1][2]

In an April 2023 interview, Mehdi Hasan challenged Taibbi over a CIS/CISA conflation and the claim that 22 million election tweets were labeled misinformation. The nonprofit Center for Internet Security and the federal Cybersecurity and Infrastructure Security Agency are different entities. Taibbi acknowledged the conflation but argued that both had real connections to election-integrity work, so the correction did not negate his broader reporting. The narrower identity error and the broader relationship claim require separate checks. [1][2]

Stanford’s own response says the Election Integrity Partnership analyzed a 22-million-tweet collection and identified 2,890 unique Twitter URLs as potentially violating platform policies. Those are different denominators: collected material, flagged URLs, and platform enforcement. Even accepting a genuine concern about pressure, a research corpus cannot simply be counted as millions of deleted posts. Stanford is an interested party, so readers should compare its explanation with the original project record. [3]

Access had its own vulnerability. On April 7, Taibbi said Twitter was restricting Substack links after Substack introduced Notes, and that he had been offered publication on Twitter instead. He chose Substack and warned this could cost future Files reporting. His first-person account illustrates why owner-granted access and editorial independence deserve ongoing scrutiny, including when the owner becomes the subject. [4]