Hey everyone, hope you’re all doing well. Let’s talk about one of the practicums for this week’s unit, the Two Headlines Bot! The Two Headlines Bot is a now inactive Twitter bot, but when it was functioning its intended purpose was to combine news headlines to produce humorous results. The Bot was created by Darius Kazemi after an exchange with a friend who sent them a tweet with the joke headline “Ben Affleck is the new CEO of Microsoft.”
Kazemi then thought of how an algorithm could be created to automatically generate other joke headlines such as that one. He decided the easiest way to do it was by utilizing the Google News Sidebar which listed the top 5 or more news stories by subject. The bot then takes the top headline from one of the topics and replaces one of the words in the headline with a different topic on the sidebar. For example, the headline “Ben Affleck to play Batman in next Superman movie” became “Ben Affleck to play iPhone 5 in next Superman Movie”.
What resulted form this bot is what Kazemi has come to call a string of “near-future late-capitalist dystpoian microfiction.” This is because Google’s algorithm for generating that news sidebar was designed to favor named entities rather than abstract concepts. So rather than the sidebar showing a concept like space travel, it would likely display the name Elon Musk, if there were news about that. The result of this is the Twitter bot Kazemi designed replacing proper nouns in a headline with a different subject. Sometimes the result doesn’t make a lot of sense, other times the headline does come out with a dystopian message Kazemi referenced, including one joke headline that read “Facebook, world powers progress in nuclear talks, agree to further meetings”.
All in all the Two Headlines Bot was a humorous account on Twitter, but it also provides a interesting glimpse at how these algorithms are designed. This bot relied on the Google News Sidebar’s algorithm which favored named entities over concepts, a deliberate choice that results in the subject of a story always being an organization, corporation, person, etc. which in turn allowed this Twitter Bot to easily interchange subjects to create humorous or dystopian headlines.
Thanks for the post! Great run down of how it works and how it ended up cultivating and finding an audience of this kind of autogenerated collision of news.
I’m curious for folks thoughts about what bots like this mean or can illustrate to use about their expressive potential for public history work. We get into this a bit with the Lubar and Mark Sample readings this week. I’m curious for thoughts from folks about how those points that Lubar and Sample brought up do or don’t connect with what Kazemi’s bot demonstrates.