Manipulated — The Cambridge Analytica Exposé — Book Notes

On Obama's 2008 online campaign: the ads placed at that time essentially amplified social validation among an already-sympathetic audience — more like customers coming to you than using data analysis to find and influence customers.

Chapter 2's title is something like "The origin of all the absurd mistakes" — the author describes their own financial circumstances and social position at the time they joined Cambridge Analytica. One line stands out: "In the past I always refused to engage with people I disagreed with, but now I'm more pragmatic. I realized that if I stop being angry at people with different positions, I can do more good for the world."

Alexander's business model: data mining to identify and reach potential clients. When vetting a client he would first ask:
1. Do you have a need?
2. Do you have a budget?
3. Do you have a timeline? — The third question reveals urgency and whether the client will actually pay.

When pitching to a Nigerian client, Alexander mocked carpet-bombing advertising (cinemas blasting Coke ads everywhere) and presented his philosophy instead: raise the cinema's temperature so the audience gets thirsty. To make people act, you create the conditions — conditions that make doing what you want them to do feel almost inevitable.

Before Facebook closed the friends-app loophole, vendors could masquerade as a friend app and harvest user data.

The methodology:
1. Segment voters in fine detail — e.g., higher neuroticism scores = higher probability of being driven by fear-based messaging.
2. Feed the voter data into algorithms to model preferences.
3. Serve targeted content on the platforms they use most — e.g., feeding desired information when specific search terms are used on Google.
4. Build canvassing apps that visually surface target profiles before a canvasser even speaks.
5. Run precision-targeted ad campaigns — more granular than A/B testing, potentially dozens of ad versions running simultaneously to different audiences.

On Brexit: the company was open to working with either side, but Remain considered it unnecessary. Leave engaged, and the description here is more detailed — harvesting Facebook user data, applying labels, segmenting, briefing clients, then precision-deploying: authentic or fabricated videos mobilizing different groups to cast the desired vote (psychological mapping of mental landscapes), ultimately pulling just enough Leave votes from targeted swing segments.

On the US presidential election: Cambridge Analytica initially wasn't working for Trump, but once he won the primary, they engaged — using precision targeting to pull a thin slice of swing-state voters to the polls.

Cambridge Analytica's lasting significance: it used fear to move people, had no ethical floor (willing to sell the same model to both sides), proved that data can reshape democracy, tore at the fabric of democratic societies, and illegally exploited Facebook data (at least according to Facebook's own statements).

In the data world, this was just the beginning. There will be a next Cambridge Analytica — only more rigorous regulation and institutional structures can prevent a replay.

One subplot worth noting: the author's relationship with Alexander showed the classic pattern of a boss who constantly paints grand visions while being consistently stingy in reality — especially around profit-sharing.