It seems to me like most things get cheaper as they get better. At least anything that benefits from improvements to a manufacturing process, economies of scale and iterated design... which is most mass manufactured goods.
Ignoring for a moment the per cycle efficiency increase (which is non-trivial and adds a couple orders of magnitude extra), a modern CPU is ~500,000 times faster than ENIAC was, which was itself a ~1,000 times faster than other computing solutions at the time, at a cost of $6,500,000 to $100.
So per dollar, you're getting 32 billion times the computer. It's actually more like a trillion times the computer if you count in per-instruction efficiency gains.
Yeah, we get better at most things, but almost no other thing have we gotten so much explosively better at it than computers.
You can argue clothes, screws, cars, shoes, whatever are better now than they used to be. Maybe they are. But they're not a billion times better per dollar. I suspect that most of them strain to be a thousand times better per dollar.
Adjusted price: $3,998
Maximum speed: 10 mph
The improvements in performance and price are not even those seen in computers.
Yes, it's much better and no I wouldn't want to go back to the state of computing in 1981, but I think that calculation speed is a poor proxy for degree of improvement.
There are arguments to be made about the first three and they might be pretty instructive arguments. One issue is whether we're looking at "what you can get for $50", "what is the best one ever made (in some respect)", "what is a typical one that people from a particular social class would buy", or still other ways of segmenting markets.