More to the point, the models that contain photons that behave “realistically” sometimes lead to unsuccessful predictions (e.g. the double-slit experiment), and models that consistently give successful predictions include photon behavior that seems “unreal” to human intuition (but corresponds to experimentally-observed reality).
All models are wrong, some models are useful.
The soft sciences use models for prediction all the time without believing that the model reflects reality.
>Why should I believe in the success of the Photon-Containing map without believing in the existence of photons?
Because doing that in the past has worked out well and led to successful predictions. You don’t actually need to assume realism.
More to the point, the models that contain photons that behave “realistically” sometimes lead to unsuccessful predictions (e.g. the double-slit experiment), and models that consistently give successful predictions include photon behavior that seems “unreal” to human intuition (but corresponds to experimentally-observed reality).