Undefeated Until It Isn't
An unbroken track record feels like proof of safety, but it usually just means the unlucky day hasn't arrived yet, and the memo warning about it was filed years ago and ignored.
Incident response, disaster recovery, and business continuity management get thrown around like synonyms, but only one of them asks what happens to the business itself when the system doesn't come back on schedule.
An unbroken track record feels like proof of safety, but it usually just means the unlucky day hasn't arrived yet, and the memo warning about it was filed years ago and ignored.
We built an entire industry around software supply chain risk in five years, then walked right past the rack of hardware where every part traces back to one factory, one continent, and a guy who's about to retire.
Security orthodoxy says patch everything, constantly. Critical infrastructure runs on decades-old systems that can't take that advice without risking a worse outage than the vulnerability it's meant to fix.
Power grids can't coordinate a recovery without their communications networks, and those networks can't run without grid power. Almost nobody plans for the day both go down together.
A small fault doesn't stay small once it dumps its load on a neighbor that was never built to carry it. Here's the plain mechanics of a cascade, and why the postmortem always sounds like it saw it coming.
A generic look at why governments keep piling resilience rules onto network operators, and why the market alone was never going to produce the boring outcome everyone actually wants.
Most of what gets called critical infrastructure isn't run by governments at all; it's owned by companies that answer to shareholders first. That mismatch between public expectation and private incentive is worth examining honestly.
Redundancy diagrams love pairs, but the failures that matter live one layer down, in the physical world nobody audits. Two "independent" feeds can share a conduit or a substation, and two independent teams can share the one person who knows how to keep things running.
In any interconnected system, overall resilience is bounded by the worst-maintained component, not the flashiest one. A tour through the usual suspects, from the ancient closet switch to the vendor integration nobody has touched since the person who set it up left.