I started close to the machine, and never entirely left.

My first years were spent writing and supporting systems that other people depended on at three in the morning. That period set a habit I still keep: before believing a report, I try to find the thing underneath it.

Over thirty years the scope changed — more countries, larger portfolios, higher stakes — but the recurring failure mode did not. Programs rarely fail because the technology was wrong. They fail because ownership was ambiguous, complexity was underestimated, or the organisation was never asked whether it could absorb the change.

I still describe myself as a student. The subjects move quickly and the honest position is that a good part of what I know will need revising. Writing here is how I do that revising in public.