What I’ve done
This workshop consisted of three main tasks, organised around INPUT → OUTPUT → PROCESS. The workshop required me to examine the data I produce (INPUT), explore how platforms interpret and categorise me (OUTPUT), and manually simulate algorithmic categorisation (PROCESS) to understand how algorithmic identities are constructed.
Task 1: INPUT — Understanding the Data Collected
Social media platforms are primarily interested in behavioural, demographic, and engagement-based data.
- They track:
- What we post
- How long we view content
- Who we interact with
- Where we are physically
- What devices we use
- Our browsing patterns even outside the app
This suggests platforms are not simply interested in our identity, but rather in our predictive value for advertising and engagement.
Task 2: OUTPUT — Questions & Example Answers
Platforms categorise users into interest-based, demographic, and behavioural categories such as:
- Age brackets
- Fashion lover, sports enthusiast, gamer, traveler
- “Likely parent,” “interested in fitness,” “tech user,” etc.
- Purchasing intention categories (“high-value spender”)
These categories are generated from your activity history.
Task 3: PROCESS — Questions & Example Answers
Datas we entered into our spreadsheet:
- The content of each friend’s last 15 posts
- Only the category it fits into
- Each post counted as 1 point
More thoughts
This workshop made me realize that most of the “data about me” on social media is not what I consciously express, but what platforms infer from my behaviour. When reviewing my data archive and ad profile, I saw not a real version of myself but a simplified identity shaped around commercial logic; the algorithm constructs interests based on traces like clicks and browsing time, and these categories often feel irrelevant to who I actually am. This isn’t simply an error, it also reflects the nature of algorithmic identity, which is dynamic, partial, and purpose-driven. Doing the manual scraping exercise also showed me how subjective and reductive any attempt to categorise social life can be, which means algorithmic bias comes not only from technical flaws but from the limits of categorisation itself. Ultimately, the workshop helped me understand that while data cannot capture a full human identity, it still strongly influences how we are seen, targeted, and interpreted online—highlighting the power structures embedded in data processing and their impact on everyday life.
Reading references (Cheney-Lippold, 2017)
- In our internetworked world, our datafied selves are tethered together, pattern analyzed, and assigned identities like ‘terrorist’ without attention to our own, historical particularities.
- “We are data” means we are made of these technical constructions, or what I describe as measurable types.
- All the while, these algorithmic interpretations are rarely known to us. They make who we are from data we likely have no idea is being used, a denial of explicit reflexivity that suppresses the “social” component of our identity’s social constructionism.
- Who we are in terms of data depends on how our data is spoken for. And in general, our data is spoken for with a language fashioned from these models.