Charts and Diagrams as Thinking Tools

Why it attracts me

I often only notice a pattern once it is drawn. A good chart is not a summary of thinking that happened somewhere else. It is where some of the thinking happens.

An example

The Process Stability Lab on this site follows a tool that tightens a bolt on every part of an assembly line, with one dot per part on a live chart. Its alarm limits are set from the first 20 parts. One of the four upsets you can trigger is tool wear: the tightness creeps up a little with every part. In a column of numbers this is almost invisible, because each reading still looks normal. On the chart, the dots start leaning to one side. Two of the lab’s alarm rules, six points in a row trending one way and nine in a row on one side of the center line, are designed to catch exactly this kind of slow change, even while every single reading is still inside the limits.

What I think (and don't know)

Pictures persuade quickly. The same power that reveals a pattern can invent one through a stretched axis or a suggestive layout. This map is itself a picture: where a dot sits on the screen reflects my choices, not a measurement.

Where it connects

Networks are both mathematical objects and visual stories (Networks: The Math of Connections). Beautiful Visuals That Stay Honest is the honesty check on everything here, and Learning by Experimenting, Not Just Reading is what happens when the viewer can change the picture.

Questions I'm still exploring

• When does an interactive chart help understanding, and when does it only feel engaging?
• How can a picture show uncertainty without overwhelming the viewer?
• What does a network drawing hide simply because of where the dots are placed?

Sources and further reading

• F. J. Anscombe, “Graphs in Statistical Analysis”, The American Statistician 27, no. 1 (1973) — Introduces the four data sets now known as Anscombe’s quartet.