Work the ratio before the outrage. Reported losses of $15.9bn against an FTC estimate near $200bn for the prior year is roughly eight cents on the dollar [3], and the agency's own gloss of $550m a day multiplies out to about $201bn, so the estimate is at least internally consistent with itself [5]. Measured against the upwards-of-$500bn figure in the AP/FRONTLINE framing, reporting captures nearer three cents [7].
Reporting is a behaviour, not a measurement, and behaviour responds to price. The FTC says its own numbers undercount because most victims are too embarrassed to file [22], and the case AP/FRONTLINE built the investigation around shows what filing buys: Simon lost $800,000 to a persona called Emily and was left repaying $185,000 he had borrowed plus tens of thousands in taxes on money he withdrew and lost [13], was told by police right away to kiss the money goodbye [14], and then had the FBI decline to comment on his case [19]. Filing had a price of zero and a cost in shame.
Which is where the liability argument gets expensive. AP/FRONTLINE's finding is that the US lags some peer countries in holding financial and social media companies responsible and in regulating the crypto rails scammers use [9], and the intuitive cost of catching up is the reported number, $15.9bn spread across payments and advertising, which for industries that size is a fee. A reimbursement duty, though, converts filing from humiliation into a claim, and the claimant population is the estimate rather than the report, about 12.6 times larger [21]. Or rather, the more interesting version: no one can hand a CFO the number until the duty exists, and a liability that only reveals its size after you accept it is a liability no legislature volunteers for. The administration and Congress are pursuing new options while victims still have little recourse [10], which is what that constraint looks like from the outside.
The counter-theses deserve their space. The $200bn estimate may simply be soft, in which case the gap is small, the duty is cheap, and the American exception is straightforward rent for banks and platforms. Or the money may be genuinely gone: AP/FRONTLINE traced Simon's funds through leaked documents and wallet addresses to a scam compound in Myanmar with thousands of victims [16], and local authorities destroyed much of that compound only for the operations to move and new sites to open immediately [17], so reimbursement would transfer cash from shareholders to victims and recover nothing from the operators. Or stigma may be inelastic: if three in ten Americans have already lost money or information [11] and still will not file, a duty changes the reported number very little.
This is probably wrong, but I read the roughly $184bn gap [21] as a measurement artefact that liability would close, which makes the reported record the least informative number in the story even as it sets a record with a 25% annual rise [1]. The falsifier is clean. Enact a reimbursement duty and watch reported losses; if they fail to double inside two years, the endogenous-reporting thesis is dead and the simpler reading, that liability was bargained away, is the better one. Erin West's line that what happens after the scam may be worse than the scam [18] is, in cash terms, Simon's loan and his tax bill.
