The Collapse of FTX

The collapse of FTX is not, at its psychological core, a story about greed. Greed is too simple - and too human. What Sam Bankman-Fried constructed was something more architecturally interesting: a belief system that pre-authorized whatever he needed to do.

Effective altruism, in its SBF variant, functioned less as an ethical framework than as a permission structure. The core logic, as he articulated it publicly and in court, was that he was earning to give at a scale that made ordinary financial and legal constraints beside the point. If you are optimizing for the greatest good, and you are the person best positioned to deliver that good, then the rules of finance, honesty, and institutional trust become obstacles rather than constraints. The framework did not cause the fraud. It provided the cognitive infrastructure that made fraud feel, to him, like arithmetic.

SBF first encountered effective altruism at MIT, where he studied physics before moving into finance. The philosophy, associated with philosophers like Peter Singer and William MacAskill, advocates directing wealth toward causes that maximize lives improved per dollar. SBF adapted this framework with a specific modification: the person doing the earning occupies a special position that exempts them from the constraints applied to others.

He described this logic openly before the collapse. "Expected value calculations are the only honest way to make decisions," he said in multiple interviews. "If you're genuinely trying to maximize good outcomes, you have to be willing to take risks other people wouldn't." The sleight of hand in this formulation - collapsing the distinction between risk-taking with one's own resources and risk-taking with customers' funds - is the terrain tell.

This is distinct from ordinary hypocrisy. The hypocrite knows the rules and breaks them. SBF appears to have genuinely believed he was operating within his framework, which is a more dangerous cognitive configuration. He was not hiding a contradiction. He had resolved it.

Caroline Ellison, CEO of Alameda Research, entered a cooperation agreement with the U.S. Department of Justice in December 2022 and testified for the prosecution at SBF's trial in October 2023. Her testimony revealed the specific mechanics of how FTX customer funds were transferred to Alameda, how losses were concealed, and how the inner circle understood what was happening.

Ellison's testimony also revealed the social architecture of the enterprise. She and SBF were romantically involved on and off for several years. The inner circle - roughly eight people, including several in overlapping romantic configurations - lived and worked together in a penthouse complex in Nassau, Bahamas. They shared an ideological vocabulary, a physical environment, and an economic interest in the continued success of the enterprise.

Ellison's journal entries, introduced as evidence at trial, show someone who understood the risks and the ethical violations with significant clarity, alongside a person who was romantically entangled with the person directing the enterprise and living inside a social structure from which exit was difficult. Her cooperation was not simply instrumental self-preservation. It appears to have been, at least partly, the act of someone who wanted to say what had happened out loud.

Between his arrest in the Bahamas in December 2022 and his trial in October 2023, SBF conducted an extensive media tour. He gave interviews to the New York Times, appeared at conferences, posted on Substack, and in general behaved like someone whose primary concern was the management of his public narrative rather than the preparation of a legal defense.

The media tour is a terrain document of unusual richness. A person who understands that they have committed fraud and is concerned about consequences does not typically give dozens of interviews explaining their reasoning. SBF's willingness to do so suggests one of two things: either he genuinely believed his framework absolved him, or he was sufficiently dissociated from the legal reality that the narrative management felt more urgent than the legal situation.

His statement to reporter David Yaffe-Bellany in November 2022, shortly after the collapse began, was revealing: "I was trying to do the right thing. I was wrong about some specific things. But the framework was correct." The framing - error within a correct framework rather than a fraud within a false one - is the self-model he brought to trial.

At trial, the prosecution established that SBF knew Alameda was using FTX customer funds, knew the funds were not available to be returned to customers, and directed the concealment of this fact from auditors and investors. The jury convicted on all seven counts in November 2023.

The question the trial testimony left open is whether SBF believed his own framework throughout. The evidence suggests a more complex picture: a person who had constructed an ideology sophisticated enough to absorb considerable evidence of wrongdoing without triggering the internal experience of wrongdoing. He could hold the knowledge of what Alameda was doing alongside the belief that he was doing good, because the framework had been built to accommodate exactly that kind of cognitive load.

He was sentenced to twenty-five years in federal prison in March 2024. The sentence reflected Judge Lewis Kaplan's finding that there was a risk of future fraud, not merely past fraud - which is its own terrain statement about the stability of the framework.

The EA community's response to FTX's collapse was largely one of disavowal: SBF had misapplied the framework, the framework itself was sound, and the fraud was a personal failing rather than a structural consequence of the ideology.

The map takes no position on EA as a philosophy. It notes only that the framework's structural feature - that sufficiently large expected value calculations justify otherwise unacceptable actions - has a known failure mode. When the person running the calculation also controls the inputs, and is also the primary beneficiary of the outputs, the framework does not constrain behavior. It licenses it. The SBF case is the clearest recent demonstration of that failure mode operating at scale.

- Lewis, Michael. Going Infinite: The Rise and Fall of a New Tycoon. W.W. Norton, 2023. - United States v. Samuel Bankman-Fried, S.D.N.Y., Case No. 22-cr-00673. Indictment and trial transcripts, 2022-2023. - SEC v. Samuel Bankman-Fried et al., Case No. 22-cv-10501 (S.D.N.Y.). SEC complaint, December 2022. - Ellison, Caroline. Plea agreement and cooperation agreement, U.S. Department of Justice, December 2022. - Ellison, Caroline. Trial testimony, United States District Court, Southern District of New York, October 2023. - Faux, Zeke. "Sam Bankman-Fried Described His Altruistic Goals. Then Came FTX." Bloomberg, November 2022. - Yaffe-Bellany, David. Coverage in The New York Times, November 2022-March 2024. - Bankman-Fried, Sam. Sentencing, S.D.N.Y., March 28, 2024.

---

Interpretive opinion based on the public record. Not a clinical assessment or diagnosis of any individual.