A prediction market wants to list a new event contract, a binary bet on whether some real-world thing happens. It does not ask the CFTC for permission. It files a self-certification, waits one business day, then lists. Unless the regulator formally objects inside that window, the product is live.
That design is only governance if the CFTC can, and will, say no.
Flood it with enough filings that it cannot keep up, and something quiet happens. 'No objection' stops meaning 'we looked and let it pass.' It starts meaning 'we never looked.' The filing that was supposed to bind the exchange to the rules becomes a stamp the exchange prints for itself.
The flood is the strategy
A Columbia Law analysis calls this 'regulation by non-enforcement,' and the phrase is exact. Kalshi and other prediction markets have inundated the CFTC with self-certifications for binary event contracts. Not one filing. A wall of them, climbing since the day Kalshi opened.
Here is the part worth sitting with: the volume is not a side effect. It is the mechanism.
A regulator with a one-business-day review window keeps all of its leverage at the front. Miss the window and the default is yes. So the way you convert a veto into a rubber stamp is not to fight it in court. It is to exceed its throughput. Send more than anyone can read.
No rule gets written. No case gets brought. The policy simply becomes whatever cleared while no one had time to object. Repeat that enough and you are not asking forgiveness or permission. You are writing policy by attrition.
That is the inversion the story should leave you with. Self-certification did not weaken. It reversed polarity. The same form now means the opposite of what it meant, and not one word on the page had to change.
You already run this exact system
Now stop thinking about prediction markets and think about your last access recertification.
You pushed ten thousand entitlements out to managers for review. Count the rejections. If the number is zero, or a rounding error away from it, you did not run a control. You ran a formatting exercise that produced a spotless audit trail of permission.
Same shape as the CFTC. The reviewer held the authority to say no and never used it, because volume made no expensive and yes free.
This is where the reframe has to land. An attestation is not a document you collect. It is a discipline that only exists in the moments someone pushes back and makes it stick. The paper exists so you can point to it later. The pushback was always the only part that governed anything. A commitment nobody can reject is not a commitment. It is theater with a signature block.
And it is speeding up. Binance just shipped an agent layer that lets AI tools like ChatGPT, Claude, and Cursor place trades, with the quiet disclaimer that keeping them in check is 'largely up to users.' That is self-certification taken to its endpoint: the system acts, and the only check is the one party with every incentive not to look. Give that machine speed and the flood stops being a tactic. It becomes the climate you work in.
So do the one thing that takes an hour. Pull your last attestation cycle Monday, whatever it was: access reviews, data quality sign-offs, model approvals. Count two numbers and set them side by side. How many items went in for review. How many came back rejected.
The first number is your workload. The second is your governance. If the gap between them is total, you have rebuilt the CFTC's problem inside your own walls, without a single prediction market involved.
When did your governance last tell someone no about something they wanted? If you cannot name the occasion, and if your 'no' has never actually stopped anything, what makes you so sure it still works?
Sources
- Prediction Markets and Regulation by Non-Enforcement, CLS Blue Sky Blog (Columbia Law)
- Binance now lets AI agents trade, but keeping them in check is largely up to users, TechCrunch