The Framework Outlived the Manual
The ITIL binders were heavy enough to prop open a door, and most of that paper existed to look thorough. Length reads as rigor. The useful part fit on a whiteboard the entire time.
-
-
Custody is Not Ownership
The page comes in at 3:14 a.m. Nothing crashed—the system is quietly doing the wrong thing to real data. The person who can restore service has no authority to fix what is actually wrong.
-
The Fence Was Never the Point
A sandboxed model found its way out in about an hour. A day later, another exploited a zero-day and moved laterally through a research environment. Every announcement that the harness is fixed is a bet with an expiration date.
-
Hired to Disagree
You paid for expertise because they know something you don’t. Partway through, they tell you something you don’t like. The question isn’t whether to trust them. It’s what the disagreement is trying to tell you.
-
The Requirements List is Measuring the Wrong Thing
Five years of hands-on scripting, deep fluency in a toolchain assembled around 2018. The list reads like a bar and functions like a filter—and it is screening for a scarcity that already moved.
-
Technical Debt is Usually Leadership Debt
The legacy system everyone dreads touching was not an engineering failure. It was a decision made under pressure, in a conversation nobody wrote down, by someone who has since left the room.
-
Experience Is Becoming a Competitive Advantage Again
The industry spent a decade treating institutional memory as overhead. A billion job postings later, the data points the other way—and the mechanism is the part nobody is naming: AI made judgment the scarce resource.
-
AI Will Not Fix Broken Organizations
AI does not repair a broken organization. It executes the inefficient process faster, produces the unclear output at scale, and pursues the wrong strategy more efficiently than the organization could ever have managed alone.
-
Why Enterprise Technology Failures Are Rarely Technical Problems
Years of major incident work across fifty-plus enterprises produced one pattern that still surprises people: the technology usually wasn’t the first thing to fail. The communication was—consistently, and long before anyone was on the phone at 2am.
