The current research in granular computing is dominated by set-theoretic models such as rough sets and fuzzy sets. By recasting the existing studies in a wider context, we propose a unified framework of granular computing. The new framework extends results obtained in the set-theoretic setting and extracts high-level common principles from a wide range of scientific disciplines. The art of granular computing for problem solving emerges from the resulting common philosophy, methodology and information processing paradigm. Granular computing stresses not only the need for rigor, structure, conciseness and clarity, but also the importance of conscious effects and wisdom in using powerful strategies and heuristics in stating and solving problems. Keywords. Granular computing triangle, structured thinking, structured problem solving, structured information processing