A small fraud does not get reported. That isn't a moral statement. It's arithmetic.
Run the numbers. Under the Commodity Futures Trading Commission's whistleblower program, an award is a slice of the monetary sanctions the agency actually collects — historically somewhere inside a 10% to 30% band, set at the agency's discretion, case by case. Now take a $200,000 token scam. A generous 20% is $40,000. Weigh that against retaliation risk, employment exposure, the privilege you waive the second you cooperate, and two years of your life spent as a deponent. The expected value goes underwater long before your finger reaches the submit button. So the tip never gets filed. The operator keeps running the book. Next cycle, the book is bigger.
Per a crypto media report, the CFTC now proposes a presumption of maximum payout on small claims — the agency would assume a top-tier award applies when the collected pool is thin, unless there's a stated reason to deviate.
That is an enormous sentence wearing the costume of a procedural footnote. The market filed it as noise. The market is wrong about which part is noise.
Context
The plumbing first. The CFTC's whistleblower program was authorized under Section 748 of Dodd-Frank. Awards are paid out of the Customer Protection Fund, funded by collected monetary sanctions — not congressional appropriation, not inflationary issuance. There is no token, no emissions schedule, no unlock cliff. The supply of rewards is a function of enforcement output, which makes it a genuinely revenue-backed incentive rather than a flywheel.
What we have is a proposed rule — NPRM territory. That means a public comment window, a commissioner vote, and a final text that can still move in either direction before it binds anyone.
Now the sourcing problem, because it caps everything downstream. The coverage I'm working from is a Tier-2 crypto-native outlet paraphrasing a filing. No Federal Register number. No comment window dates. No vote record. No definition of "small claim." No treatment of anonymous tipsters, foreign tipsters, or tipsters who were themselves participants in the conduct. No timestamp on the report at all.
I spent six weeks in late 2017 inside Tezos's on-chain governance contracts, and the lesson from that audit holds here: in early-stage rulemaking, the entire meaning of a text lives in the definitions the summary leaves out. What follows is mechanism reasoning, not a leak.
Core
Start with what the discretionary band actually was. It was never a number. It was a distribution — a range with unknown parameters, applied by an agency that retains discretion to land anywhere inside it. Traders price distributions every day; that's the job. Whistleblowers cannot. There is no order book on CFTC award ratios, no implied vol, no historical settlement series they can regress against. Which means the 10-to-30 band functioned as a lottery ticket with an undisclosed strike.
Fear is just unpriced volatility in human form. And you cannot sell a hedge to someone who doesn't know what they own.
The proposed presumption doesn't raise the ceiling. It pins the floor. On small claims, the default flips from "somewhere in the band, decided later" to "the top of the band, unless we say otherwise." For a would-be tipster, that converts an uncollateralized option into something closer to a term sheet. Stabilization fees are the tax on certainty — and here the agency is the one paying it, in future payout dollars, to buy a cleaner signal from the person holding the information.
Fine. But watch where the real cost lands.
Every enforcement agency has exactly one scarce resource, and it is not the award budget. It is investigator hours. The CFTC's Division of Enforcement runs on finite headcount — attorneys, forensic accountants, market surveillance staff. A rule that increases the volume of inbound tips does not increase that headcount by a single FTE. The audit found no bugs, but it found time. The presumption spends money to buy information, and information is only useful if someone can process it. That asymmetry is the whole story, and it's absent from every headline I've seen on this.
Everyone designing incentive rules believes they are building a mechanism. In 2020 I put $50,000 of my own capital into a Curve pool specifically to test whether the stabilizer behaved the way the paper said it would. It didn't. The oracle manipulation vector was visible in the tick data weeks before it was visible in anyone's writeup. That is the discipline this rule needs and will not get for a while: mechanism design without execution data is a hypothesis wearing a suit.
The layer beneath that one matters more. Whistleblower awards are a one-shot game. My Tezos work was a repeat game — I published a technical correction within 48 hours of mainnet launch because reputation compounds and I intended to be around for the next ten cycles. A tipster weighs a single payout against a single, permanent career cost. One-shot games with uncertain payoffs collapse; that isn't psychology, it's game theory. The CFTC is trying to convert a one-shot interaction into something that feels repeatable by removing the variance. Reasonable instinct. Harder than it looks, because you cannot make retaliation risk legible with a percentage point.
There's also a forum question nobody is asking. The SEC runs its own whistleblower program with its own award structure and its own processing speed. A tipster sitting inside a hybrid exchange doesn't file with "the government." They file with the agency whose program they can model. This rule is a bid in an inter-agency market for information, and it lands precisely where the CFTC has been fighting for jurisdictional oxygen against the SEC on digital assets. Read it as procurement and the shape makes sense.
The thing I keep coming back to is 2022. When TerraUSD broke its peg, the on-chain data showed the drain in real time — public, permissionless, free. The code screamed silence while the ledger bled. The information was never missing. What was missing was anyone with a paid incentive to assemble it into an actionable case. That gap is real, and this class of rule is aimed straight at it.
Contrarian
Here's the part that should worry the agency more than it worries the industry.
A presumption of maximum payout on small claims is a floor price on tips. Set a fixed price for a good — any good — and you change what gets supplied at the margin. The cheap stuff floods in first. Low-quality tips cost their author almost nothing to produce; high-quality tips cost insider risk, documentation, and nerve. When both clear at the same floor, you don't get more good tips. You get more tips, and the mix degrades.
That's the lemon problem, and no enforcement agency has solved it cleanly. The remedy is triage, and triage is exactly the investigator-hour sink described above. So the failure mode isn't that the rule overpays. It's that the agency pays twice — once in awards, once in staff time spent closing the ninety reports that went nowhere to find the one that didn't.
There's a compositional point the headlines miss, too. The highest-value crypto enforcement cases have come from insiders at exchanges, market makers, and listing committees — large matters, large sanction pools, comfortably inside the existing band. The small-claim presumption doesn't touch those. It targets the long tail: hundreds of modest scams, each individually rational to ignore, collectively the texture of the market. The rule changes the composition of the pipeline, not its volume. Anyone reading it as "the CFTC is about to get much more aggressive" is reading a staffing decision as a strategy.
And the definitional hole is still there. "Small claim" is undefined in everything I've read. Does it track the sanction amount, the customer loss, the number of victims? Does a $2 million case with four hundred retail victims qualify? Does a $500,000 case against a repeat offender? Those choices are the rule. Not the presumption.
Takeaway
Two things to watch, and neither is a price.
Watch for the Federal Register number and the final definition of "small claim." That single threshold tells you whether this is a broad sensor network or a narrow gesture.
Then watch the conversion data — tip volume against actions filed per enforcement attorney. If submissions multiply and case quality doesn't, the presumption was mispriced, and the bill will show up in closed matters per agent, not in the award line.
In a tape this flat, there's nothing to chase and no narrative to front-run. Execute the trade before the narrative solidifies — except here there is no trade, only positioning. The plumbing is being rebuilt while everyone stares at the chart. That's usually where the next twelve months get decided.