The full reasons for this are beyond the scope of the post; for the purposes of this post, it’s sufficient to note that slow pages are often super-linearly slow as devices get slower and that slowness on one page doesn’t strongly predict slowness on another page.

While you can get into trouble with HTML pages that are so large that browsers can’t really handle them, for pages with a normal amount of content, it generally isn’t until you have complex CSS payloads or JS that the pages start causing problems for slow devices.

Source: How web bloat impacts users with slow devices

So… how do we learn from this and move forward?