Tufte's data-ink ratio is right, and also not enough ↗

Note This is placeholder / sample text

The data-ink ratio — the idea that every drop of ink on a chart should be doing work, and anything decorative should be removed — is one of those ideas that seems obvious once stated and yet gets violated constantly.

The principle holds. The application is where it gets interesting.

The failure mode I see most often is not decoration but false precision — charts that maximize data-ink but present their signal with more confidence than the underlying data supports. You can have a very clean chart that is also misleading. Removing chartjunk is necessary but not sufficient.

The other thing Tufte underweights, perhaps because his examples are largely print, is the interactive case. A chart that rewards exploration can carry more non-data elements than a static one, because those elements earn their keep by facilitating the interaction. The guidelines shift.

Worth reading the original if you haven’t. Worth being suspicious of anyone who cites it as a complete answer.