This book breaks new ground by drawing attention to certain kinds of biases that permeate many parts of science. Data are constrained not only by limitations of measurement instruments but also by the precondition that there is some suitably positioned observer there to 'have' the data (and to build the instruments). This simple truth turns out to have wide-ranging implications for fields as diverse as cosmology, evolution theory, imperfect recall problems in game theory, theology, traffic analysis, the foundations of thermodynamics and the interpretation of quantum mechanics.
Yet, disturbing paradoxes lie in ambush. The infamous Doomsday argument is one of these, but it is merely the tip of an iceberg. By means of thought experiments and careful philosophical investigation, the book develops a precise theory, cast in a Bayesian framework, of how to reason when our evidence has an indexical component or we suspect that observation selection effects have biased our data, the theory caters to legitimate scientific needs while showing how to resolve the philosophical paradoxes. It offers new conceptual and methodological tools for thinking about the large-scale structure of the world and the place of observers within it.