Despite an increase in computer speed, software isn’t faster. It does more (the good case) or it’s simply sloppy, but that’s not necessarily a bad thing because it means it was cheaper/faster to develop.
Same with web pages. You deliver more and/or you can be sloppier in development to save dev time and money. Shaking a dependency tree for a web app, or improving the startup time for a client side app costs time. That’s time that can be spent either adding features or building another product entirely, both of which often have better ROI than making a snappier product.
ROI might say a developer should build a different product instead of speeding up the old one. Or perhaps it’s better to get 200 less satisfied customers than make the 100 existing ones more satisfied. That can be done by using the resources for marketing, features, SEO. In the end, when you are optimizing there is always something you are not doing with that time.
Whether hundreds of users value the time they gain by not waiting for page loads isn’t relevant either unless it actually converts to more sales (or some other tangible metric like growth).
Most people seem to get more confused and hesitant when pages are loaded with more features, most of which are irrelevant to their neeeds of the moment. (Of course flat design makes this hesitation worse.)
And theory talks about "cognitive overload" and "choice paralysis".
Same with web pages. You deliver more and/or you can be sloppier in development to save dev time and money. Shaking a dependency tree for a web app, or improving the startup time for a client side app costs time. That’s time that can be spent either adding features or building another product entirely, both of which often have better ROI than making a snappier product.