Joe Sabia and Aaron Rasmussen made the Webby-Nominated The Office Stare Machine after the success of The Office Time Machine. While editing together all of the references in the office in chronological order for The Office Time Machine, Joe also collected every time someone looked directly into the camera. After a few weeks of thinking about what to do with all those 706 stares into the camera, Joe and Aaron decided they'd allow users to type in any emotion and get a sequence of stares displaying that emotion.
Partly inspired by an article about how facial expressions are not interpreted universally across cultures, they immediately began researching theories on how to categorize and group emotions. They decided to develop their own grouping system by manually watching every single stare and tagging it with a primary, secondary, and tertiary emotion. Then they chose the 31 most distinct emotions from the total set (think of it as a sort of principal component analysis for feelings!). Using a list of around 500 synonyms, they linked synonyms by similarity to each of these 31 emotional keys, and voila, you can type in any emotion and get a playlist of stares that feel correct. They also ran stats on the most popular emotions people typed and looked for high rates of keys that were not in their synonym list, and then added those keys (unless they thought the searched for word was stupid). Tada! Try it. It's fun.