About
Three-part interactive work that explores how digital systems turn people into data, and how
that data is used to construct a version of us that exists within the system.
It stages
biometric extraction, behavioural accumulation, and algorithmic
synthesis as a sequential, embodied experience, culminating in a portrait built
from participants' own data.
[A]: Biometric Extraction
I focused on biometric extraction as the entry point into the system. I designed
the interaction so that the user must first wave to activate it: an intentional gesture that
mimics the subtle, almost habitual actions we perform when engaging with digital systems. Once
activated, the system detects the face in front of it, projects a field of dots onto the
surface, and begins mapping facial features into depth values.
I chose this approach to
directly reference familiar infrastructures of facial recognition, such as immigration
e-gates, Singpass verification, and Apple Face ID. These systems
rely on translating the human face into measurable data points: distances, contours, and
relational coordinates, reducing identity into something that can be stored, processed, and
authenticated. By visualising this process through projected dots and depth mapping, I wanted to
make visible what is usually hidden: the abstraction of the human body into machine-readable
information.
This first stage deliberately requires the collection of facial data to
foreground how algorithmic identities are constructed. Before any interaction can occur, the
system must first “know” the user through biometric capture. It reflects how participation in
many contemporary systems is contingent on surrendering one’s physical identity as data. In this
sense, identity is not self-defined but extracted and converted into a dataset that becomes the
foundation for how the system recognises, tracks, and ultimately defines the individual.
[B]: Behavioural Accumulation
In Part B, I explored behavioural accumulation as a way identity is formed
through interaction. I designed a system where users are presented with a deck of content cards,
swiping left to skip and right to like. While this feels simple and familiar, the system is
simultaneously tracking their behaviour, what categories they prefer or avoid, how long they
linger on each card, and their overall response patterns.
This setup intentionally
mirrors the mechanics of social media platforms, where everyday gestures like swiping, pausing,
and liking become inputs for continuous monitoring. What appears to be casual interaction is, in
fact, a structured process of data collection.
Through this, I frame datafication as
something embedded within habitual use. Each micro-decision is recorded, quantified, and
aggregated, gradually forming a behavioural profile. This profile becomes the basis of an
algorithmic identity, not something explicitly declared, but inferred through patterns of
interaction. In this stage, identity is no longer extracted from the body, but constructed from
behaviour, showing how systems come to “know” users through what they do rather than what they
say.
[C]: Identity Construction
In Part C, I explored identity construction by using the data gathered from
Parts A and B to generate a visual representation of the user’s algorithmic identity. Here, the
system translates both biometric data and behavioural patterns into a composite image. For
instance, if a user’s top interest category is politics, a larger proportion of the generated
visuals, such as 70%—will reflect political imagery.
I designed this as a way to make
visible how systems synthesise different layers of data into a singular, legible identity.
Rather than presenting identity as fixed or self-defined, this stage shows it as something
computed, assembled from probabilities, preferences, and patterns derived from user
input.
This visualisation highlights how algorithmic identities are not neutral
reflections, but constructed outputs shaped by the data users continuously feed into the system.
By externalising this process, I aim to reveal how individuals are reduced to dominant signals
and weighted categories, forming an identity that the system recognises and reinforces.
