You can’t have 1st and 2nd without the 3rd kind
#3 is not about making observations in general but about making a map of every tree. This is a common mistake of scientists, wanting to have a complete set of data on their topic. I myself have to remind myself of this as well when designing projects. But in reality you cannot make a map of every tree and even if you could, what’s it worth? You need abstractions to understand anything and models are just that. To design a good research study you need to balance how many observations are necessary vs how much effort they take.
I think the comic is about types of published articles with the 3rd being “we collated and collected a bunch of data to make this database that other researchers will find useful”-type papers. So you don’t need to rely on the map of every tree if you just map some trees yourself.