“What Would I Say?” by Pawel, Vicky, Ugne, Daniel, Harvey, Edward, Alex, and Baxter

This post is also available in: Español (Spanish)

Sample output generated from my Facebook feed.
Sample output generated from my Facebook feed.

This app uses your Facebook feed as a data set to perform a Markov Chain analysis and generate new status updates from it. In other words, it uses your status updates as a lexicon to assemble a few sentences that echo your style, interests, concerns, and topics. Written in HTML and JavaScript, this little app was created during HackPrinceton 2013, a hackathon that attracted about 500 students this past weekend (November 8-10) to create “real-world projects.” The result is this uninvasive little bot that runs on the user’s browser (client-side scripting), connects to your account with an app via the Facebook API, and posts– with your permission– the status updates it generates. The beauty of this approach is that your privacy is protected because your Facebook data and authentication information aren’t stored anywhere but in your browser and Facebook account. You can also enter a friend’s Facebook username or a celebrity’s page name and it will also generate mock status updates. Here are some examples published in the site:

whatwouldisayexamplesMarkov chains have been used extensively in recent Twitter bots, but entered mainstream hacker culture when Hugh Kenner famously published “A Travesty Engine for Micros” in Byte in its November 1984 issue, and have been available online since the mid 1990s. The output of Markov chain generators is generally legible, understandable, plausible language, but because it has no way of discerning the conceptual frames of its word choices, it results in odd and funny non sequiturs. “What Would I Say?” produces lines in an unexpected e-poem that, despite its logical oddities, carries an echo of your voice.

Featured in Genre: Bot