The Apollo virtues can improve a system while quietly hollowing out the humans inside it.
By Apollo virtues, I mean the classic operator toolkit: go to the floor, name constraints, measure flow, remove waste, repeat. I used it hard. It worked, often spectacularly. But the same pattern kept returning: output climbed, trust spent down, and people became colder, more guarded, more performative. Mechanistic improvement can succeed while humans pay.
I saw the limit on a night paving job. Trucks backed up and money burned in the dark, but the real mistake happened hours earlier, when uncertainty was high and the only socially safe move was to over-order trucks. The problem was not bad optimization. It was that optimization could not make uncertainty admissible.
That is why Ritual Time matters. Run Time can tighten a system. It cannot help a group metabolize error, status, and meaning. AI will multiply the Apollo virtues. If you do not build the other half, you will scale theater with the output.