Contradiction, in the book, is conflicting statements from the same side, each of which demands an explanation and a fact-check while the conversation moves on (p. 90). It is easy to see in a single speech. It is harder to see when the conflicting claims arrive from thousands of accounts at once. This week is about a country where someone got inside the machinery behind those accounts and came back with the receipts.
The book’s own Kenya material is from August 2022, as presidential results came in slowly. Claims that Raila Odinga had won, that William Ruto had stolen the vote, and that foreign hackers had tampered with the servers all circulated at once, and none of them was true (p. 72). That is the C variable in its plain form: incompatible stories about one event, each needing its own rebuttal. The machinery that can fill a week like that had been documented a year earlier.
In May 2021, a five-judge bench of Kenya’s High Court struck down the Building Bridges Initiative, a constitutional amendment project backed by President Uhuru Kenyatta and Raila Odinga. Within days, hashtags attacking the judges, and the lawyers and activists who had brought the case, were climbing Kenyan Twitter’s trending lists. To a casual scroller it looked like a country arguing with its court.
Two researchers working as Mozilla fellows, Odanga Madung and Brian Obilo, spent the following months documenting what was underneath: at least eleven distinct campaigns, more than twenty-three thousand tweets, roughly thirty-seven hundred participating accounts, aimed at judges, journalists, and civil-society figures. And because they interviewed the people who ran the accounts, we know the mechanics from testimony rather than from inference on network graphs.
The mechanics are the point. Participants were recruited through WhatsApp groups. On each campaign day they received a brief: the hashtag of the day, pre-written tweets to copy or lightly vary, and instructions on when to post, because the trending algorithm rewards a synchronized burst over a slow accumulation. For a day’s work they were paid roughly ten to fifteen dollars, through M-Pesa, the mobile wallet most Kenyans carry anyway.
The unit of work here was the gig, priced by the day.
The prize in every case was the same piece of real estate: Twitter’s trending column. Trending is formatted as a measurement. It reads as what Kenya is talking about. The campaigns treated it as a purchase. Buy enough synchronized posting to cross the algorithm’s threshold, and the platform itself converts your spend into the appearance of spontaneous public opinion, then displays it to everyone, including the judges and the newsrooms. That laundering step belongs to amplification more than to contradiction, and I would rather say so than stretch the case.
If the campaigns had ended there, Kenya would be one case study. What makes it a playbook is what happened next. In October 2021 the Pandora Papers named the Kenyatta family’s offshore holdings. Within days, Madung and Obilo found, the same infrastructure was running in the opposite direction: paid campaigns to discredit the reporting and defend the president, again reaching the trending lists. The first wave attacked the judiciary on behalf of power; the second defended power against journalism. The infrastructure did not care. It was a market, and a market takes clients from either side.
Twitter, to its credit, acted once the receipts were public: an internal investigation and more than one hundred account suspensions for platform manipulation. The market’s answer was to move. Ahead of the August 2022 general election, Madung documented the same commercial logic operating on TikTok: more than one hundred thirty videos across thirty-three accounts, over four million views, including content that played on the ethnic fault lines of the 2007–2008 post-election violence. Enforcement on one platform relocates the market rather than closing it.
The reason this case is worth a file, rather than a Russian or Chinese state operation, is precisely that no state was required. The popular image of an influence operation is an intelligence service with a building full of operators. The Kenya files show the retail version: a gig economy of influence, priced at a day laborer’s wage, available to any political actor with a modest budget and a WhatsApp contact. Wherever that market exists, the accounts that carry a claim, or its opposite, are available off the shelf.
The lesson for the rubric is, as usual, narrow. For contradiction, the test is the one the book gives for spotting a flooding moment: mutually inconsistent claims arriving together (p. 29). Paid campaigns add fingerprints that sincerity does not: a synchronized onset rather than an organic ramp; near-identical phrasing across accounts with no social connection; a cadence that starts and stops like a shift; and somewhere, if anyone with access looks, a payment trail. When a wave of opinion arrives, test the schedule. The sentiment is often genuine enough. Publics do not keep office hours.
Next week: A is for Amplification, and the deepfake that wasn’t.
— J.W.B.
Note 10 of 24