[Something very different this week: an app review, or perhaps an algorithm review!]
I found the first one sandwiched between a New York Times book review and a Wall Street Journal productivity hack for beating procrastination. The next day I found another, this time hidden among a pile of financial news websites still trickling out missed-the-rush Jack Bogle obituaries after his death last week. And I found another today, under a set of lifestyle articles and Trump fact-checks grouped into a quasi-advertisement encouraging me to “read more stories from the Washington Post” even though I’d already hit my monthly non-subscriber limit.
Why has anime fan fiction invaded my Google News feed?
For the past week, while scrolling through the morning headlines, I’ve been treated to a steady stream of stories from the esteemed muckrakers over at webnovel.com. For example:
The Ninja-verse’s Immortal Cultivator… The story of an oddly feminine male immortal cultivator in Naruto, and his adventures through the ninja-verse
Pokemon: Journey Towards Greatness… A war veteran and an anime fan as well, … now our MC has turned 82 years old … and waiting for death to take him away but will death be is [sic] end or will it start a new beginning, a new legend.
My Life With An Overlord And Anime System… *Ding- You have been selected to receive the Anime System. Please Think “Kono Dio Da!” to receive Hamon Beginners Pack or “Jajanken!” to receive Hunter X Hunter Beginners pack…*
Lovely little reads all, I’m sure, but do they really belong in a news feed amongst the Times, Post and Journal, Chronicle, Globe, and Tribune? Oh, what algorithmic hell hath I unleashed?
Really, I suppose it’s my own fault. Even though I don’t want fanfiction crowding out my morning papers, I’ve found each of the webnovels so funny that I click on them anyway, thus feeding whatever loop identifies me as a consumer of such content to serve me up more. And as much as I’m ashamed to admit it, Google isn’t exactly wrong: I scroll through the “For You” feed less for actual news than for the entertainment value of seeing the weird recommendations the algorithm will spit out using my online data.
I do sometimes find the increasing prevalence of data collection disconcerting for its uncanny invasion of our personal space, but I also take heart in the apparent limitations of many of the tech-giant algorithms. Though journalists like to pick up on horror stories like corporations using “predictive analytics” to identify pregnant women, or more dystopically, China’s social credit system, in my experience most consumer-facing big data applications don’t work nearly so well. Companies like Google clearly have reams of information about my life, but seem to struggle to use that data to produce fully effective online services.
The Google News feed is a prime example the technology’s immaturity. When it isn’t pitching naked adverticles like “This 55-inch 4K Roku TV is just $300,” it’s giving me comically irrelevant headlines like these:
- “The Myers-Briggs® Personality Types Of Voltron: Legendary Defender Characters” … about a web series I’ve never heard of using a silly, click-baity psychological metric of dubious scientific merit.
- “Get Religious in Elder Scrolls V with Wintersun – Faiths of Skyrim Mod” … about an obscure mod for a seven year old game that I haven’t played since high school.
- “Gemballa has Always Stood for Unrestrained Excess, and this 818-Horsepower Monster Continues the Legacy” … about I don’t even know what. Cars? What is Gemballa?
So why am I getting this stuff? I’m peering into the black-box of a tech-giant’s proprietary algorithms, but I have a few solid hunches. The Voltron piece seems easy enough to explain: Google has almost certainly – and correctly – tagged me as interested in animation (see this blog). The Skyrim mod article gets more specific: I suspect that it dates back to last weekend, when I did a flurry of Googling on modding to install an unofficial community patch for Borderlands 2. And finally, the car article likely has an immediate antecedent: I recently read a different Jalopnik piece about Japanese kei-cars after I passed by one that had slipped and tipped on an icy road.
But let’s back up and solve that fanfiction mystery. Where did all of those amateur anime novels come from?
My best guess? A couple weeks ago, a friend sent me a link to a fanfiction he found funny. I opened it in Messenger (owned by Facebook) on my Android (owned by Google), read the synopsis for a couple minutes, and returned to the conversation. But that’s all it took. With just a casual aside in a joking chat with a friend, Facebook and Google both had me tagged as a fan of fanfiction. And as a reward (or punishment?) for sharing that data, they’ve been bombarding me with articles and advertisements about it ever since…
It’s maybe a little creepy to trace the Google News recommendations back to specific events in my life (or less dramatically, specific web activity), but in general I don’t mind the data collection too much (except for security issues like mass breaches and de-anonymization). I use ad blockers on most of my devices and keep up with my own curated list of preferred news sources when I’m not just wasting time in the Google News “For You” feed. For the most part, I limit my interaction with recommendation algorithms.
However, even if I don’t feel threatened, I do feel ever so slightly manipulated. My frustration has less to do with the usual privacy concerns than the behavioral models that govern user interaction with internet-connected applications in the first place.
Giving the algorithms positive feedback is totally automatic. After all, it’s just part of the texture of online living: I click a friend’s link, search for a how-to, or do some idle reading. But negative feedback requires conscious effort – a very small amount of effort to be sure – but effort nonetheless. Even if only takes two clicks to tell Google “Fewer stories like this,” I have to interrupt the flow of my activity and stop what I’m thinking. It’s a trivial micro-interruption, but given the stubbornly low quality of Google’s recommendations, I’m not getting much of the supposed consumer benefit either. And if I’m not benefiting, why bother?
The issue falls into the realm of behavioral economics, probably best known in the media for the “nudge theory” of Nobel Prize winning economist Richard Thaler (and introduced in his own words in this simple New York Times column). At its most basic, behavioral economics studies the way small psychological tricks can result in large changes in economic behavior. For a well-studied example, consider the effect of default options for organ donation programs:
If you wanted to donate organs but your government required you to submit paperwork affirming your consent, would you care enough to go through the effort of registering? Thaler cites a 2003 study that says for most people, probably not. Though polls show that large majorities want to donate organs, in reality few do. For example, in the study’s timeframe, Germany’s opt-in organ donation system only enrolled 12% of its citizens. Meanwhile, neighboring Austria achieved a 99% participation rate using an opt-out system. Such a minor policy tweak of the default setting had the enormous benefit of increasing the supply of transplant organs for patients who needed them. But best of all, it didn’t harm conscientious objectors to organ donation, who could still opt-out of the system.
However, the organ donation “default nudge” has social welfare as its ultimate goal and preserves individual choice by allowing anyone to opt-out. By contrast, the tech giants’ automatic data collection systems focus on profit first. That isn’t necessarily bad, but at the same time, opting out of the system isn’t as simple as clicking “No, I do not want to donate my data.” Unlike Thaler’s nudges, the data defaults do limit consumer control because of the excessive complexity of just saying “No.” (Tangential aside… Thaler has another interesting column which describes private sector abuse of behavioral economics).
Take the volley of advertisements for anime figures I received after writing a blog post comparing the relative prices of Rem and Ram merchandise. Though my post only focused on physical stores in Akihabara, I also did some online window shopping out of curiosity. Of course, just like with the fanfiction, the internet’s hidden algorithms sprung into action to serve me up ads. A month later, I’m still clicking away at “Stop seeing this ad” for anime figures. Google promises that “We’ll try not to show that ad again” and I don’t doubt that they mean it. But I still sometimes see Rem’s smiling, plastic face when browsing, if not from Google, then from Amazon or some other tech giant.
To be fair to Google, they do make opting-out easy enough. I can reset the advertising ID on my Android and clear my Chrome browser cookies and work through the privacy options on their Data and Personalization page… and well, maybe I take that back. It can become a pretty difficult process for a not-so-computer-savvy consumer like myself.
This isn’t a simple behavioral nudge, like refusing to donate organs by checking “no” on a form. While writing this post, I’ve spent several hours reading privacy tutorials and muddling through Google’s settings. And even though I’m mostly satisfied with the results, I’m not finished. Google is just the most prominent company in a vast network of app developers and online advertisers. I still need to duplicate the process with Facebook, Apple, Amazon… and oh, what about third-party data collectors with less public scrutiny and thus less need to offer the same sort of robust privacy options and transparency documentation as Google?
Unless I want to become a genuine technophobe like my grandfather way out on his farmstead in the Middle-of-Nowhere, USA, it becomes a lot of work to disconnect from the great algorithmic machine while still retaining the benefits of internet use. It’s much easier to just coast along in indifference and ignorance, laughing at the stupid recommendations. Where did the anime fanfiction come from? I don’t care!
But from a different angle, I do care. I want a news-aggregation app that isn’t terrible, and over the years Google News has continuously disappointed me. I’ve played the negative feedback game and hidden dozens of junk publications like “Scary Mommy” and “Android Police” and yes, “webnovel.com” but Google seems to have no shortage of irrelevant garbage to shove through the grinder. At this point, I’m ready to give up and find a new feed.
And to do that, I’ve got to do a search on the Google Play Store…
…can’t escape the machine, eh?
Everything is Marshmallows! 