It comes form being able to tell which part of the success is because of your method and which part is the data that you fed to your method. There was a big listing of domain knowledge assumptions and this seems like a one of the first things to assume about a domain. When one knows the difference between knowledge and assumtions it isn’t that hard to take simplicity preference as an assumption instead of fact ie the proper attribution doesn’t really increase ones cognitive workload.
It can be okay to guess that simplicity works good (especially when one knows that the odds are good) but then you are not knowing.
It comes form being able to tell which part of the success is because of your method and which part is the data that you fed to your method. There was a big listing of domain knowledge assumptions and this seems like a one of the first things to assume about a domain. When one knows the difference between knowledge and assumtions it isn’t that hard to take simplicity preference as an assumption instead of fact ie the proper attribution doesn’t really increase ones cognitive workload.
It can be okay to guess that simplicity works good (especially when one knows that the odds are good) but then you are not knowing.