On August 1, 2012, Knight Capital's faulty software deployment triggered approximately 45 minutes of erroneous trading. The SEC later reported a loss of more than $460 million and found inadequate controls and written procedures.
Source: SEC — Charges Knight Capital With Violations of Market Access Rule (October 16, 2013)
I'm Kendra from The Writing Studio, and this is Paper Trail — where I go back in time to show you exactly what happens when the docs fail. Let's get into it.
What Happened
According to the SEC's order, Knight deployed a software update across eight servers, but a technician missed one. Knight had no written deployment procedures for that router and did not require a second technician to review the deployment.
Source: SEC — Knight Capital order, paragraphs 15 and 26 (PDF)
Inside their system were these things called flags — digital switches that trigger specific pieces of code. The team repurposed an old flag to activate their new software. On seven servers, it worked perfectly. On the one server that didn't get the update, that same flag woke up a retired trading algorithm called Power Peg — code that hadn't been used since 2003. And Power Peg had a broken stop counter. So it just started trading. Buying high, selling low, millions of times, with nothing telling it to stop.
Before the market opened, an internal system generated 97 automated emails identifying an error. They were not designed as system alerts, and Knight did not act on them. After the market opened, the router sent millions of erroneous orders. The SEC found that Knight lacked adequate procedures for responding to such incidents.
Source: SEC — Knight Capital order, paragraphs 17, 19, and 27 (PDF)
The Real Lesson
Here's the documentation lesson I take from this: reusing a flag while leaving old code callable creates risks that need explicit review. The SEC's findings about missing deployment procedures show why teams need a written process and a verification step, not just an assumption that every server was updated.
Written procedures are one safeguard — but they have to be paired with testing, review, and controls that can stop an error.
That is what technical writers actually do. We make sure the people who come after — new developers, new managers, new vendors — aren't flying blind. We create the paper trail that says 'this exists, here's what it does, and here's what happens if you don't account for it.'
The SEC identified multiple failures, including inadequate risk controls, code deployment procedures, and incident response. Documentation was part of that failure, not the only cause.
More than $460 million lost after approximately 45 minutes of erroneous trading. A reminder that written procedures and operational safeguards both matter.