Because the complaint isn't that it isn't testable. Since we're talking about non-testable properties of these models, the only way in which they differ is in h

Because the complaint isn't that it isn't testable. Since we're talking about non-testable properties of these models, the only way in which they differ is in how we think about them, which puts this in the realm of philosophy. And the article is arguing that the philosophical implications of MWI are problematic.