"This business of arrows is a tricky one"

—Dependency Graph for Knowledge

The best discussion I've seen on the challenge of deciding which way to draw the arrows is from Edward Tufte, in the book "Beautiful Evidence" -- and available online here: Links, Causal Arrows, Networks

He centres the discussion on this diagram (this graph, this infographic... what do we call this now? Tufte calls it a diagram and an "art chart" ! nice.)

Arrows in this diagram align with the "arrow of time" - event A caused or contributed to event B.

But thing A did not touch thing B. Thing A could never touch thing B, it was never aware of the existence of thing B, which was in the distant future.

Is it fairer to say that thing B references thing A, and to draw the arrows in the reversed order?

Data versus Control

A young man resembling myself, but younger, learned Engineering, and remembers in particular learning to create logical diagrams representing digital systems. The challenge with arrows was very evident in this subject. Often "data" would flow from A to B, but would be initiated by a "control" signal which originated at B and went to A.

Hence two types of lines were drawn in the diagram: "Data flow" -- which were "wide"/"hollow" lines, more akin to "roadways" or "pipes" in the diagram, and "control flow" which were thin lines, more akin to a single piece of "wire".

In html entities, the two arrow/line types might be best represented by:

• ⇒ '⇒' Data flow — a road or pipe carrying "a lot"
• → '→' Control flow — a wire carrying a small (but crucial) signal

References

• Edward Tufte forum: Links, Causal Arrows, Networks

See also

• Dependency Graph for Knowledge
• 7 Bridges of Königsberg