Opinion writer for Tech Crunch Nik Milanovic imagines a future in which individuals will have total control of their data identities. Total control, as he sees it, would mean that everyone would own their personal data, have it all consolidated in one place instead of fragmented across platforms, and have the ability to grant or deny access to third-party services. In terms of how such a model of digital identity could exist, Milanovic turns to blockchain technology, which is a system of information that is highly secure, decentralized, and distributed in structure. 8

Considering the way personal data is used by companies now, for advertising and personalization, it’s hard to imagine what the function of a singular data identity would be in this context, besides the ability to have control over who accesses it. While Milanovic’s vision for the future of data identity sounds idyllic, the idea that an identity, even a datafied one can be complete and cohesive across the many contexts our data is generated in, is unattainable. 

Despite the large amount of personal data that I’ve been able to download and manually consolidate, it is far from a complete model of my identity. There were plenty of gaps, missing information, and incorrect approximations. Many of the data points from different platforms and even within the same platform at times directly contradicted each other. To have a singular complete model of my digital self would be to imply that identity is singular and completely understood and can be accurately and objectively represented by data. 

The idea that identity is singular and completely understood is false both on and offline. Offline, I am understood as a different person in a classroom than I am in the TSA line at the airport or with friends. Even then, I could be understood as a certain type of friend in one social circle, and a different type in another. Identity is as much about the way we are understood by others as it is about the way we understand ourselves. It’s highly dependent on context and relationships. The same way an algorithm takes in data about my actions and uses its pre-existing framework to interpret what those actions mean, people are always trying to understand each other using prior knowledge, and there are always gaps in understanding. 

For example, if I came home from work with red puffy eyes, my roommate might ask me if I was upset about something, but the reality could be that I have bad seasonal allergies and the pollen count was high that day. We misinterpret each other all the time. Being a social human being means constantly trying to bridge those gaps in understanding. 

Gaps in understanding just take a more tangible form in data. To Google, around 3:30 in the afternoon on October 25, 2020, I was understood as a Single 18- to 24-year-old Female interested in, among about 140 other things, Apparel, Breakfast Foods, Clipart, Home Automation, and Parenting. At the same time, to Twitter, I was an English-speaking 13- to 54-year-old Female interested in, among hundreds of other things, Abba, California, Comedians, Jordan (the country), and Writing. Some of the inferences are accurate, some are too broad to be completely false, and all of them are an attempt at understanding who I am based on Google’s and Twitter’s own definitions of what an interest in Abba or Jordan or Clipart actually means. 

The mistake in accepting Milanovic’s model of an ideal complete digital identity is to accept data as somehow more objective and therefore more true than human understanding. On the idea of objective knowledge, feminist theorist Donna Haraway writes, “the codes of the world are not still, waiting to be read. The world is not raw material for humanization.” It is more valuable to understand knowledge as something rooted in your perspective, through the lens of your senses.  9 The multitude of different ways my data has been interpreted and the ways I’ve interpreted it myself have made it abundantly clear that data is not objective knowledge. In coming to terms with the question of whether I can construct a complete identity for myself out of my data, it's best to accept Haraway's concept of objectivity through partiality. You cannot have a complete, objective data identity any more than you can have a complete objective embodied, non-data identity.

The problem that distinguishes these platform’s datafied interpretations from our regular person-to-person ones is their impact on how we see the world. In 2012, Facebook released the results of a study that found that users that received informational messages about voting on their feeds had “a direct impact on political self-expression, information seeking, and real-world voting behavior.” The study illustrates how the content we see on our algorithmically curated news feeds, whether that’s Facebook or elsewhere, can significantly affect our offline behavior without our awareness. 11

The risk is that the lack of awareness puts users in a passive role. We accept the interpretations of the world that appear in our feeds as in line with our own because, based on our data, the algorithms decided that they they are. This is where the large-scale problems of misinformation and polarization come into play. 

Two studies from the Conference on Human Factors in Computing Systems found that when users were made aware of how algorithms populate their feeds or form their own theories about how it works, they alter their interactions with the platform. When there is an increase in awareness, users can exercise some control over the content they see. 10 11  Knowing what you look like as data, at least in part, and trying to recognize how that data translates to what you see when you browse recommendation-algorithm driven platforms, the experience of scrolling through your feed doesn’t have to be completely passive. You can be intentional about the data you create through your engagement with your feed. 

Having downloaded my data, I start to understand the content I see on my feeds a little better. When I see an ad on Twitter for Novartis pharmaceuticals, I now know it’s because Novartis wanted to reach an audience that looks like followers of @Atul_Gawande, which for data-related reasons that weren’t provided to me, meant that I was the audience they were trying to reach. Now when I like a tweet or scroll past a certain ad, I can guess the effect it will have on my feed and make decisions about my digital habits accordingly. 

It’s not possible to escape the reach of data collection or the influence of algorithmic recommendations unless you want to delete all your social media, throw out your cell phone, and move to a remote cabin deep in the woods to live off the land. Solutions in the form of legislation or technology like blockchain are on the horizon, but not necessarily promising. It’s hard for policy to keep up with quickly evolving technology, and I’m skeptical of a technology-reliant solution to a problem created by our reliance on technology. In the meantime, this is what I can do: take the time to be reflective and critical of the content appears on my feed and why it might be there, be intentional about the data create with my digital habits, and in that small way, I get to regain some control over my data identity.