Hi Eugene, These are very interesting ideas. We designed Graft with the assumption that users would visualize only a very very small part of the graph (e.g. a single vertex or a single vertex and its neighbors). This has a few advantages: (1) Because very few number of vertices are logged Graft's overhead is very small. (2) Visualization is simple. In order to become a complete GUI for large-graphs, we'd need to figure out how users would want to visualize large-graphs. We think that's a very interesting question but we don't know the answer. Probably not in a node-link format as even at a scale of a few hundred vertices, meaningful visualization in a node-link graph becomes very difficult.
Can you elaborate on what you mean by the logging being terse? Best, semih
