Unlike the usual situation with theoretical physics which is testable in the laboratory, in cosmological theories of the universe one faces the following problems: The observer is part of the system, the universe, and this system cannot be altered to test physical theory. Even though one can in principle consider any part of the observable universe as separate from the acts of observation, the very hypothesis of big bang implies that in the distant past, space-time regions containing current observers were part of the same system. One, therefore, faces a situation where the observer has to be considered as inherently a part of the entire system. The existence of horizons of knowledge in any cosmological view of the universe is then tantamount to inherent observational limits imposed by acts of observation and theory itself. For example, in the big bang cosmology the universe becomes opaque to radiation early on, and the images of extended distant galaxies merge for redshifts, z, of the order of a few. Moreover, in order to measure the distance of a remote galaxy to test any cosmological theory, one has to disperse its light to form a spectrum which would cause confusion with other background galaxies. Since the early universe should be described in quantum terms, it follows that the same problems regarding quantum reality and the role of the observer apply to the universe as a whole. One of the most fundamental properties of quantum theory, non-locality, may then apply equally well to the universe. Some of the problems facing big bang cosmology, like the horizon and flatness problems, may not then be preconditions on theoretical models but may instead be the manifestations of the quantum nature of the universe.