Software speed is actually a binary thing. If it's as fast as the user then it's good. If it's not then it's bad. Think of it like a car - in theory different cars have different accelerations and vmaxes, but in practice if you can reach max allowed speed in your country (except Germany) and overtake all other cars you want to overtake, then you're happy; if you cannot then you're not.

Here in hackernews we have a bubble of people who are very smart so they process information at high speed in general, and have lots of experience with computers, so they process computer-related information with extra speed. We're like race drivers, squeezing every single bit of performance from the car, while majority of use cases for a car is being stuck in traffic on your way to work. Those people simply don't care about performance above 150km/h because they'll never reach such speeds so putting expensive, high-performance engines is a waste of resources.

The point I'm trying to make is that if software is fast enough for average user, and most users are fucking slow, there's no business need to further optimize it. You might keep arguing "but but but back in my days apps were instant and now they suck" but users simply do not care. This is why companies put shitty, time-wasting animations even if nothing is being done in the background.