The construction of virtual science landscapes based on citation networks and the strategic use of the information therein shed new light on the issues of the evolution of the science system and possibilities for control. Citations seem to have a key position in the retrieval and valuation of information from scientific communication networks. Leydesdorff's approach to citation theory takes into account the dual-layered character of communication networks and the second-order nature of the science system. This perspective may help to sharpen the awareness of scientists and science policy makers for possible feedback loops within actions and activities in the science system, and probably nonlinear phenomena resulting therefrom. In this paper an additional link to geometrically oriented evolutionary theories is sketched and a specific landscape concept is used as a framework for some comments.