Here is a visualization I did few year ago. The task was to visualize probabilities of successful dice rolls for a board game. For example, wheat would be the probability to get 10+ on two dices, if it is possible to reroll one dice? Or what is the chance to get 8+ by rolling three dices and ignoring the smallest one? These probabilities are not hard to could with the computer. The goals is to present them in a readable format for quick decision making during the game.

Here is how the table looks like. Each number is a probability, expressed in percents.

So the first thing I did was to show the probabilities with the color, so that the portion of the colored background would correspond to the probability:

Detail:

Looks much more readable! But can we do better?

It is not immediately easy to tell whatever a certain probability is larger or lower then half. One have to apply one’s attention to read these values. So I made a second version:

Detail:

I split the boxes in halves, and fill these halves sequentially. The portion of the colored background still correspond to the probability, but the small datails are more visible now. I like it better!

But what if we go even further? What if instead of splitting boxes into two parts, we split them into six instead?

Now this may be a step too far. Picture now looks too noisy to me, printed numbers overlap with color bars. I submitter the previous version.