This is an exciting development. From ScienceBlog. Things are really complicated, and in order to model them *correctly* requires complex models with a lot of estimates and assumptions. Except. You can make a much simpler model that will mostly work OK. Love it.
The best part is how it applies to different kinds of models. That implies that it also applies to models that weren’t tested, like economic models. Or models of how medicines work. Or climate change models.
The key is to find the right model. Duh. At least now, we can have some confidence that it is possible.