When light is emitted by different sources, we add amplitudes (electric and magnetic fields), and intensity is proportional to the square of the amplitude. When the different sources have random relative phases, the light is said to be incoherent and the literature tells us to add intensities. The reason is that the square of a sum of amplitudes (the resulting intensity) differs from the sum of squares of amplitudes (the sum of intensities) by a quantity called the interference term, and the literature tells us that the random phases together with time-averaging makes the interference term zero. My problem is that the same arguments, that I can think of, for the interference term being zero also imply perfect destructive interference so the resulting amplitude is also zero at each point in time. I argue this by saying that the random phases over many sources means that for each contribution to the field evaluated at a given location and time, there is another contribution 180 degrees out of phase with it. Hence, there is no resulting field anywhere. Where is my mistake in concluding that random phases from many sources makes nearly perfect destructive interference? How do we explain that the time average of the interference term should be zero using an argument that does not also imply nearly perfect destructive interference?
This topic came up because I'm trying to understand various theories about decoherence in quantum mechanics via entanglement so I want to have a thorough understanding of every kind of interference that can exist and what causes them. I was disappointed to realize that I don't even have a good understanding of incoherence in classical waves so I better get that straightened out first before moving on to quantum mechanics.