no, you can deduce a new observation that is then observed. In the case of dark matter, this would most convincingly take the form of a multimodal observation.

no, you can deduce a new observation that is then observed. In the case of dark matter, this would most convincingly take the form of a multimodal observation. Predict that X particle is DM. It then has interaction Y that isn't gravitational. Then Y, which we were not looking for yet, is found.