My understanding of delayed choice quantum eraser experiments roughly corresponds to the Wikipedia article about them, so please forgive if this question is stupid.
In all delayed choice experiments I read about, the choice whether to erase the path information is made in a random fashion (e.g. using a beam splitter). Why is such randomization necessary?
What would happen if instead of making random decisions here, I would try to send information by deliberately keeping or erasing the path information of the idler photons? Wouldn't this cause the interference pattern of signal photons to disappear and reappear according to the bits of data I am trying to send?
Why cannot this method (causing the interference pattern of signal photons to disappear/reappear by deliberately observing or erasing the path information of idler photons) be used for superluminal communications?