One takeaway was that given a large sample of sites, the baseline size correlates fairly well with how the bundle size scales as projects grow. So if you have a 65kb baseline, you're more likely to end up with a larger bundle size in a "real world" project than if you used a framework with a 5kb baseline. The reasons are not clear (they could be anywhere from direct causation to indirect causes like incidental complexity due to some communities' higher likelihood of using libraries without auditing them), but the results are what they are, nonetheless.
Also, 65kb gzipped means you're running the uncompressed amount of JS. As a rule of thumb, the more JS you're running, the slower the site is going to be. This matters a lot for mobile, where JS execution speed can be absolutely abysmal (on top of slow network speeds, higher chance of resource timeouts, etc).
[1] https://timkadlec.com/remembers/2020-04-21-the-cost-of-javas...
It's also 65kB of compressed and minified JS which ends up being about 185kB for doing absolutely _nothing_.
For comparison the complete jQuery library weights about 30kB compressed and minified.
It’s not like the 65kb here is 1kb for every N characters of text. It’s an overly simplified “hello world” example. The point is that you’ll still see sizes of around 65kb even with much more complexity than “hello world.”