measurement

Perfect digibodies

Quietly, we are all accumulating 4D digital doppelgängers, representations of ourselves in the data we leave behind when we have our actions measured and tracked. What's at stake, and how can we think about these traces?

Proportions of the human face, by G. Audran. Image in the Public Domain, via Wikimedia Commons https://commons.wikimedia.org/wiki/File:Bowles%27s_Proportions_of_the_human_body_Fleuron_N015596-34.png

For probably a decade and a half, I’ve been interested in and concerned with technologies of the body. I’ve been interested in the way human bodies are represented and accommodated, how they’re measured, and how individuals try to navigate a world designed and built on the basis of extreme simplifications and generalizations. My doctoral dissertation was mostly about these questions, and how they relate to digital capture and fabrication technologies like 3D scanning and 3D printing. When I turned my dissertation into a book, I doubled down on the body emphasis and replaced less-corporeal case studies with ones about mass-customization of clothes and shoes. Basically, the question of how human bodies are accommodated (or not) within a changing technological, social, and economic landscape has been on my mind for some time.

For a while, I was trying to make a term happen for the growth of data collection about bodies (and behaviours). The term was “perfect digibodies,” riffing on the digibody, a concept from Mary Flanagan. Flanagan's (2004) articulation of the digibody was as a kind of human-shaped embodiment of data – at the time Flanagan wrote about digibodies, there was a proliferation of digital characters online, meant to emulate and potentially replace humans as personalities in interactive worlds, or in one well-publicized example, as newsreaders. Flanagan made the argument that these data-based simulacra of mostly women reproduced the inequalities, norms, and biases already present in the way people in general, and women in particular are policed in the world. The digibodies were not a revolutionary new form, but a media product with the same old expectations of what people (mostly women) should be, look like, and do. Flanagan’s digibodies were mostly constructed characters, springing from the minds, desires, and market research of the media industry.

Your 4D doppelgänger

Ten years after Flanagan presented the digibody as an idea, I became obsessed with the question of how an increase in tracking of body data would change the way humans understand themselves and are understood by others. The idea of what I called the “perfect digibody” was that it would be not a constructed character, a digital embodiment of an imaginary human, but instead a collection of real and real-time, constantly augmented data about a real person. You would have your own perfect digibody, a duplicate of yourself. In the early 2010s, this was already beginning to happen. The collection of data about things like physical activity, even heart rate, was already underway. The first uses in legal proceedings of data from fitness trackers were taking place. Insurance companies were offering clients cheaper premiums in exchange for access to data from those fitness trackers. It felt, at the time, as if we were entering a world in which our vital statistics would soon be constantly monitored, recorded, consolidated, and used to build a yet-more granular picture of our lifestyles, health, and the space we took up in the physical world – not always for our benefit. Essentially, I saw Flanagan’s idea of the digibody, alongside developments in health and body tracking, and thought that the next step would be constant surveillance of our bodily existence.

I wasn’t wrong about the underlying trend, but the way it’s happened has ended up looking very boring. That’s often the case. Distressing developments that would be a big deal if they were obvious and looked novel get to slide in by the side door when they’re baked into boring interfaces that provide some marginal utility. Big changes are hidden in small things.

When I wrote about perfect digibodies a little over a decade ago, I didn’t manage to put together a whole argument, or anything long enough or detailed enough to publish. It was always an in-progress project, sidelined by things that were more immediately important. But one thing I wrote back then still looks good to me, and that’s the definition I came to:

[A] composite digital representation of your body, used in your absence as a model of you, your history and your potential. Your perfect digibody, the representation of you based on a huge, rich set of data about your contours, your inner workings, your behaviour, all contribute to the use and analysis of this digital surrogate. But the perfect digibody itself is an artefact over time that maps and tracks your body and its changes in minute detail. It's perfect because it's not just a snapshot in time, but is instead a personal, individual, longitudinal model of you. It's not just traces, it's increasingly everything. It's you in four dimensions. And every exciting new product that tells you something new about yourself, your body, your health, contributes to the enrichment of that perfect digibody and to a more detailed picture of you, in the past, in the present, and in the future.

That still feels like a good definition of a perfect digibody. The wild proliferation of methods by which human bodies, their characteristics, and their activities are tracked has only added to the possibilities for putting different data streams together to create detailed digital doubles of humans.

To make it less abstract, the kinds of data and connections I meant could be (among very many other data points) basic things like how many steps you take in a day, and where you’ve been. Your phone has this information now, most likely, and especially if you use its health features. This is data that, twenty years ago, could have existed, but wasn’t always happening automatically, simply as an artefact of having a phone in your pocket. Even ten years ago, these things weren’t so automatic, but they were coming. The ease with which tracking now happens is astounding, if you compare it to how data collection used to happen.

What's happened since 2014

In 2014 when I was first trying to articulate the idea of the perfect digibody, I was also tracking my heart rate during bike rides. This involved having a little fitness tracker on my wrist, which was paired to a heart-rate band worn around my chest, and I could get charts of my data by physically connecting the fitness tracker to my computer and transferring the data. No location information was correlated with my heart rate data, because getting a fitness tracker with that feature would’ve been more expensive, and my phone, which might’ve provided the relevant information, wasn’t involved in the process at all. Now, wearing a smart watch would be enough to achieve what, a decade ago, I had to deliberately buy a product for. This is before we even start thinking about things that are now basic features of life, like being able to view data about the duration and quality of our sleep.

Aside from the ease with which tracking can now happen, where are we today with the development of perfect digibodies? A shorter, sharper definition I wrote back when I was first thinking about the concept goes like this:

A perfect digibody, as I constitute the term, is what happens when you routinely submit yourself to biometric, anthropometric and behavioural capture and analysis. It's the composite body made up of all of the data about your body, its functioning and its trajectory over time, held by others.

Those last three words are where we can find one of the few crumbs of good news. Much of the data we generate isn’t being correlated with the other data, but is instead in a variety of different silos. Even data that would be easy to correlate, like location and heart rate, because those are both being tracked on personal devices located within the same smartphone ecosystem, isn’t inherently being put to use for the benefit of the company providing the service. Possibly because it would end up being a legal nightmare otherwise, Android’s health data platform, for example, stores data locally, on your own device, instead of being housed on a Google server somewhere. It’s not foolproof, but it’s something, at least.

I’m painting a bit of a rosy picture of the situation. An enormous amount of behavioural data is collected about individuals through their use of online platforms – this isn’t directly information about the state of your body, but it’s become very clear that data about behaviour can be used to make inferences about, for example, health. This behavioural data is not handled with the same care as the health data housed on our own devices. It’s even collected without our active consent, and without being done with our benefit in mind. Beyond our digital traces (and the reason I’m coming back to perfect digibodies now, ten years after I was previously working on the idea), the increased digitization of everything, but especially government services, means that the connectability of data is that much easier. Fragmentation still stands between where we are now and the risk of everyone having a perfect, granular, trackable, digital double. Mercifully. But trusting to fragmentation isn’t a long-term strategy for avoiding the risks presented by having too-accurate data-based doppelgängers. It doesn’t take a lot for that fragmented data to be connected – the huge data grab in the United States led by DOGE is one example of this, with personal data that was previously fragmented and stored by different departments, with walls in between, becoming centralized.

When humans are data

Everything we do is increasingly a data stream. We're data when our vital signs are taken and tracked in a medical charting system. We're data when we communicate over platforms that store everything we've said. We're data when we purchase things and those purchases are logged, either in aggregate, or individually. To take a term from Shoshana Zuboff, our lives are increasingly informated, with actions and attributes being recorded and tracked as we go about our daily lives. All of this informating means that basic activities are now producers of data. Listening to music on a streaming service like Spotify produces a trail of information, as does watching videos on YouTube – we know about these ones. But your actions and movements in your house can also be producers of data, if you own a smart TV or networked thermostat. The array of objects and actions that can be thought of as digitally-mediated has been expanding at a hugely rapid pace, and continues to do so. A string of breaches and outages in internet-connected objects seems to not be enough of a wake-up call for the general public to see data and information as artefacts of not just screen-based media, but as exceptionally valuable traces produced by our least actions. It doesn’t help that the companies trying to sell “AI” solutions are very interested in real-world data for their training sets, knowing that building a model of the world works better when there’s more real-world information. There's a new data land grab happening, and it's about recording what's happening in the physical world.

So what’s the value of perfect digibodies as a concept? Hilariously, the value is in the way a number of diverse streams of argument and knowledge get consolidated by the idea of the perfect digibody. There’s some irony for you – a concept about aggregation and consolidation is useful as a consolidator of disparate things. The way we generate data through our very engagement with day-to-day goods and services, the way that data is cross-referenced and used by others, the way we may benefit from the data, and the way different forms of data are protected and regulated, are all present in the idea of the perfect digibody. Differing from terms like “digital twin,” which offers a neutral-to-positive image of data doubles (and which still has a much stronger association with things and cities than with people), the idea of the perfect digibody comes with a bit of menace and a bit of caution. The data we collect and have collected about ourselves may offer us some benefits, but it also opens the door for enormous risk.