Personalization: the Swimsuit Edition

Last week, I decided to look for a swimsuit online. Spring is coming, summer... I checked a few sites. Pretty quickly came to the realization that regardless of whether I wear my old suites or buy new ones, I would be disappointed with the outcome. (I have given up on chasing the perfect body and decided to be a writer instead, after all.) So what was the point? I abandoned my carts and continued wasting time on the internet.

Since then, on every site I go, the yellow bikini and retro one-pieces await me. Stretched over adolescent torsos that have less fat than my left forearm, they interrupt my Facebook feed, welcome me on CNN, and follow me on The Guardian. I am considering online shopping for something innocent, say, floral patterned teapots, just so I don’t have to look at girls’ bodies all the time.

The whole experience brought back something that I have been thinking about for a while. With so much data about what we like and what we want out there, and not much scruples on accessing it, combining it, monetizing it, are ads the best we can do? Rather than eliminating friction so we can consume more of the stuff we are already gorging on, could we not direct consumption towards improvement? And how do we do this without relinquishing agency and control to the algorithms that would be calculating not what we like but what would make us better?

Improving Ourselves

What started it all was When in the late 90s Amazon opened its online bookstore, it changed the game of retail forever. Netflix and user generated content platforms followed. From watching the same must-see TV and reading the bestseller’s list, we suddenly had access to a much wider pool of content. We became excited by the idea of the long tail. We all became content creators and distributors. But above all, we became content consumers caught in the hamster wheel of the stuff we like and agree with, unable to get off even if when we get exhausted and depleted.

So how can the concept of personalization be applied if our goal is intellectual improvement?

  • The augmentation: What if personalization could challenge us to evolve our knowledge? What if instead of feeding me the same trashy thrillers I like to read on vacation, Amazon nudges me to evolve my taste? “You have been reading thrillers and mysteries set during WWII. Here is a history book recommendation that will make your knowledge of WWII more comprehensive.”

  • The provocation: Inadvertently, we have let the big three, Apple, Google and Facebook, filter our exposure to news. As a result, we lament the eco chambers we have created for ourselves, but do nothing to open them up. What if your news platform intentionally and openly introduces you to different opinions? “You have been reading The Guardian and The New York Times. You should consider balancing your current affairs reading by checking out The Wall Street Journal or The Weekly Standard.”

  • The long term goal: I want to write. But I often worry that I do not have anything interesting to say and am easily distracted. So every time I open my computer determined to write, I waste time on the Internet. What if instead of stalking ads, I get reminders that I should be writing, that the time I spent on CNN or Facebook could have been used for achieving whatever productivity goal I had set for myself that day?

Improving Our Relationships

Facebook knows a lot about me. It is really good at bringing up random potential connections and nudging me to become friends with them on the platform. But once I do, it does not really care if and how I interact with them (outside of birthdays and self-referential Facebook friend-iversaries). Gmail can finish my sentences, making responding to an email (and questioning the sincerity of any communication) as easy as clicking on a button. Slack supposedly can discern gender bias in communication, but since the company announced it can do it more than a year ago, it has gone silent.

With so much data available and the algorithms to process it, should we not be able to do more than just predict ad relevancy? What if the email client goes beyond suggesting platitudes but warns you that the tone of your email is too abrupt for the situation? What if your social network thought about the quality, not quantity, of your relationships and suggested ways to make them more real? Given how cavalier they have been about accessing our phones and emails they could at least say: “It has been six months since you emailed, messaged or phoned Jane. You should reach out to her…”

The Feather and the Chicken

Being constantly reminded that you can use your time and your mind better can get pretty old, pretty quickly. It can also be quite creepy, making you question your agency and ability to make autonomous decisions. The reality, however, is that we are already being manipulated. The outcome of this manipulation, however, benefits only data-hungry, ad revenue-driven companies. If we are already giving up so much of our data and our privacy, should we not demand more?

The line between voluntary improvement and free will-imprisonment is delicate. To paraphrase the wise Terry Pratchett, it is the difference between using a feather and using a chicken. If we are to think of personalization as a way of improving ourselves and our relationships, how do we design meaningful personalized experiences while avoiding chicken territory ?

  1. Give the user the ability to declare intent. We operate in different modes. Sometimes I want to waste time, sometimes I want to be productive or learn. Yet, none of the online platforms you can lose yourself into allow you to declare intent. “Hey Browser, I am in a learning mood. Help me learn about the history of Sicily and make sure I am not distracted in the next two hours.”

  2. Be transparent about the outcome the system is steering you towards and the rationale for that direction. Behavioral studies show that we are likely to put up with the pain if we can envision the gain. That’s why some fitness, health and financial services apps not only set measurable goals but encourage you to visualize them — saving is easier if you have the image of the car you want to buy or home improvement project to finance in front of you. Similarly, Apple News can organize its feed not based on things you read, but based on things you should read to have a balanced view. “This news is recommended to you because you have been reading one-sided, extreme arguments.”

  3. Give the user control over the parameters. At all times users should be in control over which tools can and should deploy personalization, as well as be able to turn off, pause and redact the recommendations the system surfaces to them.

  4. Enable the ultimate personalization option: the option to pay for what you consume, without being tracked. Let’s be realistic.

As I am writing this, I am not entirely convinced that any of these suggestions will be an improvement at all. In fact, they may be even more dangerous, as their manipulation is hidden behind a more noble intent than “Buy!” Would I be better off if my browser introduced me to fitness classes or shut down automatically to make me go for a run, instead of chasing after me with swimsuit ads? Or maybe the experience will be so off-putting that I would shut it down and do something else. Of my own free will.