Byrne Hobart - FTX, Drugs, Twitter, Taiwan, & Monasticism

Dwarkesh Patel opens with FTX. Byrne Hobart first warns that much is unknowable: many decisions ran through auto-deleting encrypted chat and the accounting is missing, so some gaps may never be filled.

He sets out two poles. In the fraud version, Alameda was a Ponzi scheme from the start and the exchange was launched to bring in cash to repay old debts. In the maximally exculpatory version, which he says is still really bad, Alameda was a real trading firm, FTX cut corners while raising money, and poor internal accounting let the combined entity lose track of whose money was where, so Alameda spent customer money.

The distinction matters because the remedies differ. Fraud is hard to fix since many people are willing to lie; bad accounting is more solvable through real auditors and consistently audited proof of reserves. He adds that a smaller fraud would not make the collapse acceptable.

His own estimate is an 80 to 90% chance of some fraud, but probably late and as a desperate move. He frames it as a thought experiment: on discovering that over a billion dollars of customer money had been spent, the ideal response is to confess and step down, while the realistic one may be to scramble, which backs a person into fraud after earlier incompetence.

frauds have to be good at accounting

A fraudster needs real books plus fake books that tie out. His MF Global example: customer funds were wired out just before the nightly wire cutoff, followed by an email blaming an error, which bought one night of liquidity. FTX's reported slips, such as burning USDC by sending it to a nonexistent address, point away from that level of sophistication.

Patel says that after researching FTX and interviewing SBF for about an hour, he and many others saw him as the most competent genius in finance. The reality was lost billions, spreadsheets thrown around and emojis on Slack payment requests.

Hobart offers two explanations. First, performance: SBF may have played up an eccentric-genius image, as in the anecdote of refusing a haircut because he had to look kind of crazy. He compares this to people who copy Steve Jobs by imitating the easy part, being obnoxious, but not the hard part.

Second, narrow excellence. At Jane Street, a trader finding ETF price discrepancies need not know how counterparties are vetted or how liquidity backups work. A firm that aims to keep trading when markets break needs such contingencies, for example buying deep out-of-the-money puts from a counterparty that will actually pay.