This document discusses graph visualization techniques. It begins by explaining that data must be visually encoded to be understood. Different types of data, such as quantitative, ordinal, and nominal data, are best visualized using different encodings like position, length, area, hue, and saturation. Common graph layouts include nodes and edges, matrices, edge bundling, and hive plots. The document outlines common mistakes to avoid, such as using 3D, poor color choice, lack of labels or legends, and not including tooltips or interactivity. It promotes starting with the question, meaningful encoding, interaction, filters, aggregation, and using high-quality tools to untangle complex graphs.