The layout algorithm does everything without any user intervention, except perhaps the choice of the layout algorithm to be used. Sometimes a set of rules can be coded to choose automatically (and dynamically) the most appropriate layout algorithm for the particular type of graph being laid out.

The end user is free to improve the result of the automatic layout procedure by hand. At times the end user can move and “pin” nodes at a desired location and perform the layout again. In other cases, a part of the graph is automatically set as “read-only” and the end user can modify the rest of the layout.

The layout algorithm is completely redone (“from scratch”) each time the graph is changed.

When the layout algorithm is performed a second time on a modified graph, it tries to preserve the stability of the layout as much as possible. The layout is not performed again from scratch. The layout algorithm also tries to economize CPU time by using the previous layout as an initial solution. Some layout algorithms and layout styles are incremental by nature. For others, incremental layout may be impossible.