Most observational techniques in astronomy can be understood as exploiting the various forms of the first-order correlation function g (1) . As demonstrated by the Narrabri stellar intensity interferometer back in the 1960s by Hanbury Brown & Twiss, the first experiment to measure the second-order correlation function g (2) , light can carry more information than simply its intensity, spectrum, and polarization. Since this experiment, theoretical and laboratory studies of non-classical properties of light have become a very active field of research, called quantum optics. Despite the variety of results in this field, astrophysics remained focused essentially on first-order coherence. In this paper, we study the possibility that quantum properties of light could be observed in cosmic sources. We provide the basic mathematical ingredients about the first and the second order correlation functions, applied to the modern context of astronomical observations. We aim at replacing the Hanbury Brown & Twiss experiment in this context, and present two fundamental limitations of an intensity interferometer: the requirement of a chaotic light source and the rapid decrease of the amount of correlated fluctuations with the surface temperature. The first of these limitations paradoxically emphasizes that the exploitation of g (2) is richer than what a modern intensity interferometer could bring and is particularly interesting for sources of nonthermal light. We also discuss new photon-counting avalanche photodiodes currently being developed in Grenoble, and their impact on limiting magnitudes of an intensity interferometer. We conclude by briefly presenting why microquasars in our galaxy and their extragalactic parents can represent an excellent first target in the optical/near-infrared where to observe nonthermal light and to test the use of g (2) in astrophysical sources.