> It could be argued that some serious science starts with "just looking at correlations"

Sure, but it can't end with "just looking at correlations". The rigorous experimental follow through has to be there otherwise its just data analytics not science. Sure analyzing data is a key part of science but its not even close to all of it.

Sure, intution is necessary, but they just took a bunch of data from all over the world and fit it to curves. This particular appoach hasn't born a lot of fruit; at least give us some natural experiments.

Just feels like makework. And even if the data show that, e.g., killing off all hummingbirds would improve pollination rates, no one would be like "yee-haw let's do it", right?