We will go a step forward from the Obama Post. Remember in that case I counted all the words in more than one hundred Obama speeches.
Using the same data I made a new algorithm based on a simple tree structure but, in this case, the branches are the words.
The tree structure is highly random, but the size o the words (branch width) is always proportional to the number of times the word appears in the speeches.
The image contains four samples with different color variations. The branches take their color from the original Obama photo post. I added some transparency so the overlap generates more color variations.
The algorithm generates a new random version of the tree every time I click. So I just clicked till I obtained a nice one and then saved it as pdf.