Interesting comparison, if a bit arbitrary. It raises a couple of questions though.
1) How do the numbers come out when you exclude images?
It's valid and good to know the total sizes, including images, but that can hide huge discrepancies in the experienced performance of a site.
For example, a page with 150KB of HTML/CSS/JS and a single 2.1MB hero image can feel very different from a page with 2MB of HTML/CSS/JS and a few 50KB images.
If we're just interested in total bandwidth consumption, then sure, total size is a good metric. If we're interested in how a user experiences the web, there's a lot of variability and nuance buried in that.
2) What device and methodology were used to take the measurements?
In this age of responsive design, CSS media queries, and infinite scrolling/deferred loading, it really matters how you measure and what you use to measure.
For example, if I load a page on my large retina screen and scroll to the bottom, many sites will send far more data than if I load them on my phone and don't scroll past the fold.
I only skimmed the article and didn't dig in to the references. These questions may be answered elsewhere.
Ah, very cool, thanks for sharing. I'm familiar with a lot of the tools for evaluating the performance of a single site (e.g., that I'm developing), but I'm pretty ignorant of the standard approaches for these types of larger scale benchmarking projects.
I actually wasn't aware that the site could also perform the test, but the Page Speed Insights Chrome extension is one of my go-to performance tools. The output looks the same, though formatted a little better in some places. The extension is nice though, because it lets you check private sites.
1) How do the numbers come out when you exclude images?
It's valid and good to know the total sizes, including images, but that can hide huge discrepancies in the experienced performance of a site.
For example, a page with 150KB of HTML/CSS/JS and a single 2.1MB hero image can feel very different from a page with 2MB of HTML/CSS/JS and a few 50KB images.
If we're just interested in total bandwidth consumption, then sure, total size is a good metric. If we're interested in how a user experiences the web, there's a lot of variability and nuance buried in that.
2) What device and methodology were used to take the measurements?
In this age of responsive design, CSS media queries, and infinite scrolling/deferred loading, it really matters how you measure and what you use to measure.
For example, if I load a page on my large retina screen and scroll to the bottom, many sites will send far more data than if I load them on my phone and don't scroll past the fold.
I only skimmed the article and didn't dig in to the references. These questions may be answered elsewhere.