I’m a 26-year-old from Montreal, a 27-year-old in Vegas, and a retiree with a penchant for photography. Mostly I’m unemployed. I’m a gamer, a fashionista, an occasional beer drinker and a dog owner. I graduated from a community college, I like to travel, and I wear glasses. Also, I might be Jewish.
Last week, I paid 10 strangers $5 to write character studies of me based on the advertisements that show up in my web browser. I told them nothing about my age, gender, location, lifestyle or career — instead they were asked to stitch a biography together from banner ads for bicycles, underwear, software, books and movies. While no one managed to Sherlock out my life story from 2000 advertisements, the results offer an interesting look at a kind of second self that we all carry around with us: Our browser history doppelgängers.
“The random individual who viewed web-based ads over a two week period is most likely an unemployed man in his early thirties.”
The data for this experiment was gathered using Floodwatch, a browser extension I’ve been working on at The Office for Creative Research, with Ellery Royston, Ian Ardouin-Fumat and Ashkan Soltani. Floodwatch was just released today, which means that everyone can easily track and analyse their browser-based ad histories. This gives us a chance to see something that is mostly hidden from us, the aggregate of browser activity and behavioural histories that make up our individual advertising profiles.
Floodwatch lets users visualise their ad history. Ads can be sorted by various criteria such as category, colour and publisher.
Although many of the assumptions that were made about me based on my ad history were wrong, few were groundless. I’m not Jewish. However, we’ve been working on a project for the Museum of Jewish History so I’ve been visiting a lot of websites for Jewish cultural organisations. I don’t live in Los Angeles, but I was there for a conference, and did do a lot of local LA browsing. I’m not a big gamer, but I’ve been following the politics of the gaming community fairly closely over the last few months.
“Some of the advertisements look like pornography so it is probably safe to assume that he lives alone.”
Many of the ads in my Floodwatch history are artifacts from these algorithmic leaps of logic, evidence of the blunt-instrument approaches to profiling that are used in web-based advertising. They can also be reminders of the long memory of our ad identities — that brief dalliance with a jewelry site 18 months ago means I’m still seeing ads for watches every day. The aggregated portrait is not so much an image of me as it is an image of what advertisers see of me. It is an identity assembled from fleeting glimpses, long-expired actions, and, more than anything, guesswork.
Floodwatch allows users to view visualizations based on their unique advertising history.
It is also an identity upon which decisions are being made. The ‘net worth’ of my aggregated ad data is assessed every time a new page is loaded, and individual ads are targeted directly to me, meaning that I see a different set of advertisements than you would, when visiting the exact same sites. In this process advertisers are not much different than my Mechanical Turk biographers, scraping together bits & pieces of my history to try to make a convincing profile. This profile is then used to decide which advertisements I see, and which ones I don’t.
“…he seems to be fun, exciting and kind with a bit of an extravagant lifestyle… just a little lonely is all.”
As the FTC’s Chief Technologist Latanya Sweeney recently wrote, “consumer differentiation is at the heart of online advertising.” What is most worrying is the ease in which this differentiation can turn into discrimination. We know for a fact that people are being shown different ads depending on demographics like age, gender, perceived income level and geographic location. Does this constitute discrimination? What kind of impact do these practices have on marginalized groups?
In order to investigate these questions, we need data. The algorithms and techniques that advertisers are using are largely proprietary and almost always opaque. Without large scale records of who is seeing which ads and when, it is extremely difficult to understand what is going on; much less to gather enough evidence to persuade (or legislate) advertisers to change their tactics.
“Although a dreamer with hopes and aspirations of grandeur, John needs to focus on his current situation and build his future on reality, not dreams.”
Our goal with Floodwatch is to provide a way for individuals to see and understand the data that is being gathered about them by online advertisers. Are the algorithmic biographies being written about you accurate? Are they fair? Are they dangerous?
At the same time, we’re seeking to gather the world’s largest collection of advertising data, and to make it available to researchers and investigators. Armed with this information, we can better understand what the advertising industry is up to: who is competing for your eyeballs and how they are doing it. Together, we can develop strong strategies for combating discriminatory practices.
You can download and install Floodwatch here.
Jer Thorp is a software artist, writer and educator living in Brooklyn. He is a co-founder of The Office for Creative Research & adjunct professor at ITP. From 2010-2012 he was the Data Artist in Residence at the New York Times. His post originally appeared on Medium.