Whose Fault Is It, Really? A journey through fragmentation, blame, and performance

Fragmented. If there’s one word that fits the UK railways, it’s that.

We’ve got a lot of moving parts:

• Network Rail keeps the tracks and signalling running.
• Train Operating Companies move fare‑paying passengers.
• Freight operators shift aggregates, containers, petroleum.
• Rolling stock companies lease the trains.
• Ticketing agents sell the journeys.
• Government bodies set policy.
• Regulators, safety boards and the British Transport Police keep order.

So when things go wrong, who do we call? Usually the name on the train. But a broken rail or failed signal isn’t the operator’s responsibility. It still hits them—angry passengers, missed connections, punctuality targets slipping—and often compensation to pay. Whether they can recover that upstream is another story; the punctuality score sticks.

Network Rail may point to operator behaviour creating repeat faults. Operators point back at infrastructure. And around we go.

It’s not just trains.

• Mobile: we blame the network for poor data while a dozen background apps chew through bandwidth.
• Home broadband: buffering must be the ISP throttling us—never mind the kids gaming upstairs and a smart home saturating the router.
• Enterprise apps: “It’s the firewall/AV/auth/VPN.” The default defence? Eliminate everyone else first—then come back to us.

In other words: prove it’s us. Not helpful. We pay for a service; when we suspect an issue, the burden shouldn’t sit solely on the customer. We can’t switch off production protections to “test properly”. And yes, it all “works in their lab”—labs rarely mirror the real world.

With the engineering talent in these organisations, surely we can do better. Build shared observability: a minimal, agreed set of traces and high‑level telemetry passing between parties. Enough to pinpoint the troublesome hop without exposing crown jewels. Enough to turn finger‑pointing into problem‑solving.

The Promised Cloud didn’t end fragmentation; it multiplied it

Cloud and SaaS were meant to simplify this—“don’t worry, it just works.” In practice, they add another layer: your ISP, someone else’s cloud, another provider’s edge. Uptime SLAs are fine, but they don’t guarantee performance where users actually feel it.

Fragmentation won’t disappear. Coordination can. Design for joint accountability—shared signals, transparent hand‑offs, and user‑centred performance targets—and we’ll spend less time proving blame and more time fixing the right thing, fast