Tables Not Graphs
The best data visualization is text and a small table.
Graphs are pretty and easy to "interpret" but very difficult (if not impossible) to interpret well.
Data analysis is ultimately about making decisions on the basis of the analysis. A graph can plot hundreds of data points. You'd never have a table with hundreds of rows as your output -- how would you make a decision from that? The truth is that you can't make a decision with hundreds of data points in a graph either, but graphs hide the problem. When people make decisions from graphs, they mentally summarize features of the graph and use a low-dimensional set of features to make a decision.
Our argument is that those features could instead be written directly into a small table. Essentially, graph readers replicate the process of finding a small set of decision-relevant parameters, but in a way that is not replicable or transparent and requires a lot of work from the reader.
We believe that decision-relevant parameters should be explicitly specified. Doing so makes the decision-making process more transparent and easier to scale across an organization.
Tables force the analysis to synthesize the data to a small number of outputs, enabling faster decision-making. It's not impossible to misinterpret a table, but it is harder to do so because the numbers are what they appear to be. That's why we think they are a superior data visualization to graphs and charts, no matter how glossy.