The micropayment economy, and the limits that make it safe

Graduate · enrolled learners

This lesson opens with The Flash Crash, 6 May 2010.

What happened

On 6 May 2010, US stock markets fell about nine percent and recovered within roughly twenty minutes, a swing of around a trillion dollars in market value that happened at machine speed. A large automated sell order hit a market already under stress, and high-frequency trading algorithms reacted to each other in a rapid feedback loop, selling into falling prices faster than any human could follow, so some stocks briefly traded at absurd values, a few near a penny, others near a hundred thousand dollars, before the cascade reversed. The loss is recorded as zero because the market largely recovered within the day and many aberrant trades were cancelled, but that near-recovery obscures the lesson: for those minutes there was no adequate mechanism to halt machine-speed activity that was feeding on itself, and the crash ran to its extreme before it stopped on its own. Regulators responded by building circuit breakers and limit mechanisms that pause trading when prices move too far too fast, precisely because automated activity at machine speed cannot be governed by humans reacting in real time; it has to be bounded by limits set in advance.

The decision point

Automated activity at machine speed can cascade faster than any human can intervene, so it cannot be made safe by human oversight in the moment; it has to be bounded by limits set in advance. The Flash Crash is the demonstration from trading: algorithms reacting to each other drove a trillion-dollar swing in minutes, and the fix was not faster humans but circuit breakers, automatic limits that halt activity when it moves too far too fast. The machine-to-machine micropayment economy, autonomous agents paying each other and paying for services at high frequency, has exactly this property: agents transact at machine speed, in volume, and a bug, a feedback loop, or a compromise can run a great many payments before a human notices, and those payments are fast and final. So the limits that make such a system safe are the same in kind as the circuit breakers built after the Flash Crash: spending caps that bound how much an agent can pay over a period, rate limits that bound how fast, and kill switches that halt the agent when thresholds are breached, all set in advance because there is no time to set them during a runaway. The decision when building an agentic payment system is therefore to treat these limits not as optional guardrails but as preconditions, because a system of autonomous senders at machine speed without them is a Flash Crash waiting to happen in payments, where the trades cannot be cancelled afterward.

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.