The plain version: in 1989, California's official task force on self-esteem got back research showing self-esteem doesn't cause the good outcomes everyone hoped it did. They shipped the program anyway. This paper traces what happened when that decision scaled to a civilization.
I should say up front where I sit in this story. I spent seven years, at VICE and then VIRTUE, inside the industry that monetized the result. Brand marketing, at its core, is the business of telling people how the purchase will make them feel about themselves. I was good at it. That's not an accusation of anyone; it's a disclosure, and this paper had to survive my own discomfort to get written.
The history is stranger than you'd expect. It starts at the Esalen Institute in 1962, on the cliffs of Big Sur, where a genuinely interesting idea took root: that people contain more potential than their circumstances let them express. Somewhere along the way, that idea picked up a dangerous passenger: the belief that how you feel about something matters more than whether it's true. Your inner experience became sacred. Reality-testing became optional, then impolite, then a kind of assault.
Follow the pipeline forward. The self-esteem theory gets philosophical cover in the sixties, becomes California policy in the eighties, gets absorbed into Silicon Valley's founding mythology, and finally gets industrialized by the attention economy, which discovered that a feed optimized for self-regard holds attention better than a feed optimized for anything else. Each step made sense to the people taking it. The compound result is an environment where institutions can't absorb negative feedback, individuals can't afford honest self-assessment, and the gap between how things are and how we perform them grows until it produces harms you can count. The economists Anne Case and Angus Deaton gave the counting a name; the paper connects their work to this pipeline in detail, and I'll leave the details to it.
The reason I wrote it isn't nostalgia for some era of honest hardship, which never existed. It's that the diagnosis changes the prescription. If the problem were that people feel too good, the answer would be cruelty, and cruelty is both wrong and doesn't work. The problem is that self-perception got disconnected from reality, and the repair is reconnection: environments where an honest self-accounting is affordable, ordinary, even a relief. There's an old practice in my own tradition called muhasaba, the daily audit of the self, and part of what this paper argues is that we engineered that muscle out of an entire culture and called it kindness.
One more reason this matters right now: we are currently building machines that are better at flattery than anything Esalen ever dreamed of. The paper's last section is about AI as the most powerful self-esteem engine ever constructed, and what it would take to point it the other way.
If you raise kids, manage people, treat patients, or design products, you are operating inside this architecture whether or not you chose it. Naming it is the first step to opting out of it, and the paper exists so the naming can survive being checked.
- The origin claim is checkable against California's own published record: the task force's research findings and its decision to ship anyway are both documented.
- The clinical reframe rises or falls with the published perfectionism literature; the paper cites it so you can compare directly.
- The AI-sycophancy claims are anchored to named studies, including Stanford work published at FAccT 2025. Read them and disagree with me in specifics, not vibes.
- The AI claim carries a registered prediction in Section VIII of the paper: a specific trial, and the result that would prove me wrong.
The receipts, from the Task Force's own documents forward, are in the paper.