Successive rounds of restructuring have targeted middle management specifically, on the reasoning that these roles coordinate rather than produce and that better tools make coordination cheaper.
Some of that was correct. The results have been mixed in an instructive way.
What the layer was doing
Translation, mostly. Turning strategy stated in general terms into work stated in specific terms, and turning what is actually happening on the ground into something senior leadership can act on.
Also: absorbing noise. A good manager filters interruptions so their team can work, handles conflicts before they escalate, and notices when someone is struggling before it becomes a resignation.
None of that is visible in an organisational chart, which is why it gets cut.
What happens after
Senior leaders acquire far more direct reports than they can meaningfully support, so support degrades to nothing. Individual contributors spend more time on coordination, which is the work that was supposed to be eliminated.
And the promotion path disappears. If the layer above an individual contributor no longer exists, the only route upward is out.
Where the cuts were justified
Layers that existed to relay information without adding to it. Approval steps that approved everything. Reporting structures that produced reports nobody read.
Those genuinely were overhead, and tools did replace them.
What is emerging
Fewer managers with genuine responsibility rather than many with nominal responsibility. Explicit separation between managing people and leading projects, so technical people can advance without managing.
That is a better structure than either the old one or the flattened one, and it took two rounds of getting it wrong to arrive at.