I think these control rooms were superior in some respects to modern software system observability.

- modelling the system rather than implementation (system status rather than many individual service statuses)

- supporting causal reasoning: the control flow on top means you can trace failure modes back, visually; software systems typically only model their own ontology, and you need to look somewhere else for the next abstraction down

- surface state first rather than time series; a pretty graph is nice to look at, but for actionability sometimes what you need is the flashing red light

- prioritize first-out indicator. In a complex system with lots of alerts, the most important diagnostic alert is often the first one - the rest are downstream and contribute to alert fatigue, despite them probably being more important business metrics

These older systems design principles have really scratched a part of my brain and I'm keen to keep pulling that thread. Do you have any recommended readings on the topic?