A rough thought on whose responsibility it is when a chart gets misread.

They are not as bad as people say. They are bad in a specific, avoidable way.

A legend makes the reader do extra work on every single data point. Direct labeling is almost always the better trade.

A chart without annotations is a question without an answer. The label is where the reasoning lives.

Most audiences can read a bar chart. Far fewer can read a violin plot. The gap between what analysts reach for and what audiences can parse is larger than we admit.

Tufte's case for showing many simple charts instead of one complex one holds up better than almost anything else he wrote.

The bottleneck in most data organizations is not computation. It is communication. The ratio of engineers to writers reflects a misdiagnosis.

Numbers don't carry their own meaning. Context is not decoration — it is the substrate that makes the number interpretable at all.

A reference post demonstrating every supported formatting element — headings, lists, blockquotes, code, and more.