We’re agreeing. Excel produced massive value because it accepted catastrophic failures. That’s my point.

The problem isn’t Excel. It’s trying to get Excel’s accessibility in infrastructure whilst demanding engineering reliability. You cannot have both. Kubernetes won’t accept Excel-style disasters, so it still needs specialists; now specialists who must learn the abstraction and the fundamentals.

You’re right: work not done by specialists often wouldn’t happen at all. That’s the choice. Accept Excel-esque failures for democratisation, or accept expertise is required.

My point is that currently available tools promise both, deliver neither.