Joel Simon

28 Jan 2014

Provocative [SFW] although many of the links are not sfw.

So Pornhub has lots of data, specifically data that is not discussed in public and data that has a few interesting trends.  I’m really not sure to make of some of these trends, like breakdown of time on site by state. Or which TV events reduced porn watchage the most (It was the Walking Dead finale). Other interesting facts are that christmas reduces por watching on average by 22 percent, except Japan where it rises by 8. I wish they had more interactive elements and the ability to see the distribution of arbitrary terms.

Well Made

This is a breakdown of popular words in inaugural speeches from Obamas second to Nixons first. Each word is given a size and position based off of its prevalence. Highlighting a word reveals words use din conjunction to it. Each word can have a color chart showing wither distribution by party, time or category.


Chris Harrison (from CMU) used a data set generated by no other than Aaron Swartz of all 735,323 books on Amazon at the time. The books are represented by a giant graph in which edges are similarities. Running a clustering algorithm generates an image of each book colored by amazon category.