"At a high level, most of the systems I mentioned learn about us through our behavior. Some of that behavior is subtle, like how long we linger, what we skip, when we hesitate. Some of it feels more intentional like likes, follows, comments, subscriptions etc. In most cases, we're not carefully declaring who we want to be to our products."
"Over time, we learn that these implicit behaviors function as levers. If we want less of something, we skip it. If we want more, we engage. We adapt ourselves to these janky systems. It's a very narrow kind of control because we don't know the weight of each actions on our experience. Then once a year, we're handed a polished summary of who the system thinks we are. But there are better ways to explore this."
Algorithms infer personal profiles from both implicit behaviors (dwell time, skips, hesitation) and explicit actions (likes, follows, subscriptions). These behavioral signals become levers that users manipulate—skipping to reduce exposure or engaging to increase it—yet users lack visibility into how actions are weighted. Annual, polished summaries present algorithmic identities that can feel personal but provoke unease. Transparent, legible feedback that returns structured data, distributions across dimensions (such as political spectrum), and potential blind spots can enable more meaningful reflection and control. Design that reveals underlying signals and uncertainties supports informed user agency over algorithmic portraits.
Read at Medium
Unable to calculate read time
Collection
[
|
...
]