Bad visualization practice plagues many infographics, degrading the legibility of the data and harming the reputation of the responsible groups. Sometimes there are happy accidents, however. Even bad practice can still end up displaying information accurately on occasion. Here are some of those instances where things line up just right.
3D pie charts are notoriously bad, but there are two datasets (100% and 50%-50%) that actually work as 3D pie charts when they are at the right orientation.
Rainbow color scales are horrible for displaying data, but sometimes the data range is such that the result ends up on one end of the scale. Subsets of the rainbow color scale often create great color scales, and this results in visualizations that are not too bad perceptually.
Exceeding the maximum number of categories for a visualization can be detrimental for a visualization, but when some of those categories total to 0%, using a visualization that would normally be maxed out isn’t such a bad thing.
These examples are exceptions to the rules. Most bad practice results in visualizations that are not the optimal way to communicate information, and often end up being more confusing than anything else. Be careful about implementing good practice, and don’t rely on lucky happenstance to solve your visualization problems.