For midlife-crisis reasons too embarrassing to explain, I recently entered an online charity sweepstakes for one of those tricked-out RV vans that young people live out of while rock-climbing and surfing and doing glamorous young-people things.
These vans represent a lifestyle and subculture known as #VanLife, and after spending an hour clicking through videos and websites about it, I decided that van life looked pretty good to me. I confirmed the sweepstakes was legit, concluded that a lottery was my only chance of affording one of these vans, and fired off my charitable donation.
Evidently, by sending money to this sweepstakes, I tripped some DEFCON 1-level klaxon alert set by the internet’s advertising algorithms, those unsettling bots that monitor your online consumer behavior and serve up customized ads. For the next three weeks, I was swarmed with ads for #VanLife-adjacent products and services: camping equipment, portable solar power systems, 12-volt heaters, mobile WiFi devices. Facebook gave me ads for motorcycle and RV insurance. Random websites piped in listings for travel services, Airbnb rentals, and BF Goodrich tires.
Getting caught out in my van life fantasy made me feel conspicuous and vaguely embarrassed. How much do these adbots really know about me? How fast do they learn?
I turned to my preferred technique when overly caffeinated: a scientific experiment! I would create an online alter ego with the intention of tricking the adbots. By introducing a new consumer profile into the algorithmic ecosystem, I could see what kind of ads I got in return. I would watch the watchers!
Thinking fast, I made a new user account on my laptop, downloaded a fresh browser, and created my alter ego: “Aunt Fiona,” a composite character based on the crazy Scottish aunts I grew up with. I gave Aunt Fiona a specific list of interests, starting with real-life details: Exotic brandies and whiskeys. My aunts were all deeply dedicated to weird, syrupy liquors. I also decided that Aunt Fiona, like my real-life relations, would be interested in old Hollywood movies and left-wing UK politics. I added a curveball: 1980s punk rock.
As a civilian, I could not match wits with the adbots. They’re too wily, too relentless.
I spent a very fun afternoon on my new Aunt Fiona account, clicking around to pursue her various interests: The Ramones, Labour Party candidates in Glasgow, apricot brandy. I wondered what the adbots would make of my odd combination of consumer behaviors.
Alas, I never found out. Google suspended my account after a few hours, presumably because security algorithms had flagged my account as clearly fake. It made sense. I wasn’t acting like a real person online. I wasn’t sending or receiving emails. I wasn’t buying anything. I imagine people try to create alter egos all the time, for purposes of trolling or identity theft, and Google flags them as a matter of course.
Google’s security systems had shut down my half-assed attempt at science, but in a way, it confirmed my suspicions. I was being watched, constantly. It seemed like everything online was tracking my engagement with everything else online.
This isn’t far from the mark, says Christo Wilson, a professor of computer science at Northeastern University. I had called to see if he could help me figure out another way to mess with the adbots. As it happens, Wilson and his fellow researchers have been deploying similar experiments to track the trackers.
“We use this technique in some of our work; we call it personas,” Wilson says. “We start with a clean browser and then we have that browser engage in very specific activities to signal some kind of behavior or demographic. One persona might be interested in jewelry, so it goes to a bunch of jewelry websites. Or this one over here likes guns; it’s a gun nut. This one over here is into women’s fashion…” (The researchers also occasionally ran into trouble with Google suspending persona accounts, Wilson says.)
Wilson and his team then documented the subsequent ads delivered to the personas, to study the link between the behavior exhibited and the ads. Their findings basically confirm what we have all experienced: Google and other players in the ad-exchange business routinely swap profile information to deliver targeted, relevant ads to online shoppers. They generate this information, and deliver the ads, by tracking pretty much everything we do online.
“Google is the biggest tracker online, and they also host the biggest advertising marketplace,” Wilson says. “They actually make lists of the interests that they’ve inferred about you, available for you to see and modify.”
Now here was some news I could use. So long as you’re signed into your own Google account, you can directly access this list of your own inferred interests, in plain English. Following Wilson’s instructions, I drilled down through my Google account settings to view the natural-language tags associated with my user profile. Try it yourself: Click the user portrait in the top-right corner of any Google webpage or app, then click through Google Account / Privacy and Personalization / Ad Settings.
I was stunned. Over the years, Google has generated a list of 156 things (I counted) I am apparently interested in. Some of the tags were spot-on (Baseball, Science Fiction Films). Others were puzzling (Multi-Level Marketing and Business Opportunities). Others were just bizarre (DJ Resources and Equipment). With a little digging, I spotted the tags generated by my sweepstakes adventure: Trucks, Vans and SUVs; Autos & Vehicles; Travel & Transportation…
I tried turning the tags off, one by one, to see if I could start fresh and reseed my consumer profile with some more interesting choices — 15th-century Andalusian architecture, say, or druidic sex rituals — something to mess with the bots. But I was foiled again. Every time I refreshed the page, the tags that I’d switched off switched back on. It was remarkably infuriating. (A simpler fix: after following the path to Google’s privacy settings, I could turn “ad personalization” off.)
I backed out and clicked a link marked My Google Activity, which lists more recent notes on my behavior that haven’t been sorted and tagged yet. Scrolling down before me was a list of all the websites I’d recently visited while researching algorithms that track all the websites I’ve recently visited. Spiraling, spiraling, into madness!
I surrendered. It was clear that, as a civilian, I could not match wits with the adbots. They’re too wily, too relentless. I decided to cut my losses and deploy the only sure-fire anti-surveillance technology I’ve been able to master. I replaced the piece of electrical tape I keep over my laptop camera. I don’t trust anyone anymore.
If you’re in the market for something less adhesive, Wilson recommends flipping that “ad personalization” off-switch in Chrome or switching to a browser that blocks tracking by default, such as Firefox, Brave, or the DuckDuckGo Privacy Browser. He also notes that all the major browsers support third-party extensions that block tracking and advertising. His personal favorites are uBlock Origin and Privacy Badger.
The bitter little coda to all of this: I didn’t win the sweepstakes. My extremely American urge to get something for nothing led to nowhere at all. But I felt like I took a kind of journey anyway. I found out what Google thinks I’m interested in purchasing, which seems like it should be useful somehow.
I’m tempted to bequeath my retail profile to my kids, when I pass, so that they can continue shopping in my honor. My immortal soul will someday depart this vale of tears, but my purchasing potential will live on.