Idea 1: A lot of my research for grad school has looked at privacy. I’ve thought for a while that it might be interesting to make a privacy-invasive robot. Imagine a robot that started talking to you and said, “Hi, my first name is Roboto, what’s your first name?” “My name is Golan.” “Nice to meet you Golan. Have you ever been to Robotland?” “No.” “Oh well, that’s where I’m from. Where are you from?” “Pittsburgh.” “Hey, the land of black and gold… bzzzz.” “Did you know that robots are German? My name is Roboto Merkle. What’s your last name” “Levin.” At this point, the robot would have more than enough information to scrape the web for as much information as it can about you, and it does so. It also takes a sinister turn and starts presenting your information to you. e.g. “Golan, I’m happy to meet you. Will you be my friend? I see you already have 4,431 Facebook friends. What’s one more? Oh, and this is a nice picture of you that I just found in my database. Hehehe.” I think this piece would make a point about how much information is available online, ready to be correlated with only a modicum of information to start with.

Idea 2: More particularly, a lot of my research this year has focused on online behavioral advertising, the practice of advertisers profiling a user based on the websites he or she visits. There is already a tool that came out a month or two ago that enables the visualization of information of who’s tracking you: the Collusion add-on for Mozilla Firefox. The idea of collusion is that websites are attached in a graph to advertisers, letting you see which advertisers know about which sites you’ve been to:

However, I think the weakness of this tool is that you can’t actually see what these sites could infer about you, which is the important part. It’s one thing to know that the HuffPost tracks you on sites X, Y, Z, but a tool that pulls out the articles you read, the information contained in them, and let’s you know when you’ve crossed academically validated thresholds for losing your anonymity would show more clearly how non-PII is becoming the new PII.