But what is the actual web usage of an individual? How do you define that? The problem is you can't. People browse and read at different speeds, click on different things, open different numbers of tabs, visit different websites and webapps, etc. It is entirely subjective. I'd argue that this idea of the average web usage of an individual is actually even less representative of reality because of its subjectivity. Every one has their own personal experience, so which one of them is real? How do you even measure and compare these results?

If you wanted to build a fast car, you would optimize every aspect of the car piecemeal. You isolate the engine by itself, improve its fuel efficiency, RPM, and horsepower. Similarly, you isolate the body and chassis of the car, optimize it for less drag and lighter weight. Eventually you do this for all the different components and they come together to make an objectively fast car. If you started with the idea of "an average driving experience", you'd end up with the typical use-case of going to the grocery store and picking up some food. Then EVERY car is considered fast. That IMHO is relatively meaningless.

Likewise, benchmarks help breakdown the problem of making a fast browser into smaller problems that can be tackled in a meaningful way. A lot of these smaller problems being solved or made faster come together to craft a truly fast browsing experience.

In example people have different configurations and addons, which has a huge impact on performance. Synthetic Benchmarks are often done with default configurations.

Then the argument here is that benchmarks should INCLUDE different configuration and addon permutations in their testing, not that benchmarks are useless.