Optics is attractive for computing because of its distinct advantages over electronics, in particular the inherent non-interaction of multiple beams passing through, or near, each other and the potentially massive parallelism. In order to compete with electronics these two properties should be fully exploited. Therefore it is of importance to consider what fundamental limitations will apply to the density of optical sources, or secondary sources, in terms of the acceptable levels of cross talk that will occur in an adjacent array of detecting elements. In order to beat electronics the dimensions of each decision making plane must be kept as small as possible. The physical consequences of this necessity must be well understood before such questions as optimal architectural structure can be addressed.