Not only that, but they are also served differently. For example, the React example has an "F" grade for static caching, whereas some other examples have "A".

Not only that, but they are also served differently. For example, the React example has an "F" grade for static caching, whereas some other examples have "A". Doesn't sound like something a JavaScript library can influence. Why is it that way? Because these are just sample projects from different people created at different points in time, served on different hosts, but suddenly brought under a single umbrella as if it was an organized project from the beginning. There are big differences in build systems, e.g. some examples are optimized by hand to inline critical CSS at the build level, others aren't. There's nothing that prevents applying the same strategies to all examples, but the fact is some example authors had time to do it, and others didn't. Wouldn't it make sense to at least use the same build system for Preact and React examples? I appreciate the effort by Addy and other folks from Google to highlight performance issues in web apps, but as it stands right now I consider the website to be rather misleading.