For my media diary, I built a Chrome extension that scrapes the DOM for every tweet in my newsfeed, and ingests them into a growing dataset.

Parsing the DOM was challenging, because Twitter has intentionally made it very difficult for people to access and control their own timelines. Class and ID names are encrypted, and there is a hairball of nested <div> and <span> tags.

Nevertheless, I was able to extract 1) the screen_name of who posted the tweet, 2) if the tweet was promoted, 3) the domain and url of the link in the tweet if it had one, and 4) a rough approximation of the tweet text.

Unfortunately, due to some database problems, I was only able to scrape 357 unique tweets over the course of the week. Of these, 15.2% where ads.

To start, I did some exploratory word clouds of the words used in those tweets:

And then moved on to look at only the promoted tweets

Next, I looked at the top accounts I was exposed to

Mostly academics, which makes sense given how I use Twitter.

Mostly sparkling water? This was very puzzling to me.

Finally, I looked at all the domains from links I was exposed to

These mostly correspond to media platforms and news channels.

Overall, I was surprised how unsophisticated the ads I was exposed to where. Only on account, Explorium_AI, seemed to personalize an ad to my profile. The rest seem very generic.

Other than the ads which are bizarre, the rest of the content I encountered makes lots of sense given the accounts I follow and media I enjoy to consume.

Now that I’ve got the pipeline set up, I’m excited to keep the experiment running and update this periodically with that I’ve discovered.

Phone:

• Health data — steps taken, miles walked, flights of stairs taken
• Map data — GPS location
• Photos — photo of several flowers, a few dogs, screenshots of a few articles I read
• Browser — Read the news (NYT, Slate, Nymag, New Yorker), browsed Twitter newsfeeds, liked a number of Tweets, browsed Instagram (mostly dog videos)
• Texts — text conversations I had with a few friends where I sent them the screenshots I took from a few news articles
• Messaging apps — Sent messages on FB Messenger, WeChat, Signal
• Overall — Roaming data, connected to home wifi and used 4G, battery life data

Laptop

• Microsoft Word — Wrote a few lines of a poem
• Browser — purchased a pillow on Amazon

iPad

• Watched a few episodes of show on Hulu

Transportation

• Swiped bus pass twice

Services

• Swiped in at the YMCA gym
• Entered library but did not swipe library card (but not sure if presence was recorded in some other way?)

Purchasing

• Swiped QR code at Whole Foods for Amazon membership, pretty sure all my motions are recorded the moment I step into a Whole Foods
• Credit card data — CVS visit, Whole Foods visit, online Amazon purchase

Appliances

• Smart fridge — I don’t think our fridge has Internet connectivity but it does record things such as the amount of ice left and when the filter needs to be replaced

In 2020, the screens in our pockets serve many roles in parallel. They are our mailboxes, our cameras, our lifelines. For these important functions, and plentitude more, they have become inseparable from us – a necessity for modern life. Yet when I went without my own for 24 hours I found myself not missing these critical functionalities. For in addition to serving the functional roles of organizing and processing the social information around us, our screens act as cartographic beacons, helping us navigate the temporal and spatial eddies of life.  Our screens are timekeepers, calenders and maps. These uses, with simple substitutes (e.g. watch, notebook, intuition), feel less salient than the variable rewards and addictive nature of social media, email and texting. But on that sunny Presidents day I realized those were the features that I missed the most. I made the accommodations I knew I needed to make it through. 

I wore my watch, wrote down my appointments the day before, and intentionally made no plans for the morning – knowing my lack of an alarm clock would make any pre-10-am ventures futile. Yet as I went about my day, I could shake the feeling that something was off – something was missing. What is my watch was running slow? What if I wrote down the wrong place for a meeting? My screen had been my ground truth. I could also double check it to verify information, and that act of checking has come to bring a sense of recenteredness and control. Without my screen, without the knowledge that I could check if I wanted to, I was floating. Perhaps it was this sense of being untethered, or perhaps it was the fact that it was the first warm and sunny day in Boston in over a month, but it felt liberating. 

I was surprised by how much psychological influence my screen exerts on me. Even if inactive or off in my back pocket, its potentiality lingers. It has grown to be a part of me, an extension of my senses. Thus to sever this connection felt like a mild, socio-cognitive flavor of phantom limb syndrome. An ant on a log. I liked this. It felt novel and refreshing and uncomfortable. Now, I appreciate my screen more, as a useful and fun bundle of information. And in this appreciation, its implicit embeddedness into my life and body has been denaturalized. 

Addiction is far too complex a term to cleanly map to the life of a modern information cyborg. In some aspects, it fits. We are dependent on our screens, and do experience discomfort when they are taken away. Yet by that definition, we are also addicted to friends, food and clothes. Other definitions attempt to square away addition with harm and negative outcomes. But these notions are too prescriptive for the complex way screens impact our lives. We may indeed enjoy the chronology of the phoneless, as I did, but it is by no means a better way to live absolutely. 

Testing