Nice straw man. Did anyone ever try to say that it's not worth fixing a 1000x-slowdown error because "we'll just buy 1000 times the hardware, that's cheaper!"?
Obviously there is a balance somewhere between pointless, unmaintainable, expensive over-optimisation on one end and programmer incompetence, bad tool choices and ignorance of hardware limitations on the other.
There will be a sweet spot in the middle, possibly different for all projects, which just has to be figured out by experienced people who knows what they're doing. Why all this trying to take "sides"? It's like getting all fired up about which is "better", planes or ships. They are both good for different purposes, choose the one that fits your need!
Obviously there is a balance somewhere between pointless, unmaintainable, expensive over-optimisation on one end and programmer incompetence, bad tool choices and ignorance of hardware limitations on the other.
There will be a sweet spot in the middle, possibly different for all projects, which just has to be figured out by experienced people who knows what they're doing. Why all this trying to take "sides"? It's like getting all fired up about which is "better", planes or ships. They are both good for different purposes, choose the one that fits your need!