Comment by wnewman

"If you look at a detailed diagram of a complex design, it is even worse than text."

Not necessarily. (Or horses for courses, or YMMV, or however you want to put it.) For some kinds of things, diagrams seem to do awfully well. Not all kinds of things, there are plenty of people who overuse diagrams, but some kinds of things.

E.g., consider the success of Feynman diagrams. They're at the edge of my expertise (I did a Ph. D. thesis on QM calculations, but not the kind that uses them) so I'm not 100% confident of this, but I have the strong impression that no one has developed a textual form that represents those relationships in a way that most people find comparably clear.

Or consider the humble "graph" (in the common usage, not the math "graph theory" usage). I do not want to deal with the text replacement of a nontrivial scatter plot in a typical experimental paper, or a diagram of a clever waveform in an radar ECM monograph.

Or consider the humble map (again in ordinary usage, not math usage) of e.g. Florida or the US interstate highway system.

Electronic circuit diagrams, medium-complexity Venn diagrams, and probabilistic inference networks also seem to be cases where diagrams can be hard to beat.