Recommender systems, demystified
Engagement signals. Watch-time optimisation. Implicit feedback loops. The actual machinery that decides what you see, and what gets denied visibility.
Depending on how chronically online you are, do you know that your favourite social media app is designed to influence your digital behaviour and the experiences you have online? What you post, what you read, the content you consume, how you feel, the interactions you see, what goes into your personal feed; these are all shaped by deliberate design choices that are engineered to keep you hooked.
An initiative to not only have you understand how platform design plays a role in amplifying harmful content, but to reimagine how social media could look if it were designed for social cohesion and well-being.
Harmful content does not spread on its own. Feeds, rankings and reward buttons are engineered choices, and those choices decide what gets amplified.
Design can work the other way too. Platforms can be engineered for social cohesion and well-being, and some of that work has already started.
The way in is reading. We sorted every resource on this page by how much of the system you already see, not by age or expertise. Find the sentence that sounds like you, and start there.
You use social media every day. You constantly get emotionally stirred: tired after a long scroll, jealous of someone's post, shaken by a comment thread.
But you have not really asked why those feelings happen, or who decided how the apps work.
You probably assume the For You Page is random. You probably assume the toxic stuff online is just because of the people who produce the content and those who encourage them.
From "I never thought about it" → to seeing the design.
You're able to name what is happening to you and to others on social media. You recognise specific harms (hate, mis- and disinformation, doomscrolling, body image strain, polarisation) and you no longer accept the premise "that's just how social media is" as a sufficient answer. You're beginning to feel critical about who is responsible, and how the architecture of social media is engineered for specific values.
From seeing the design → to pushing on it.
You understand that platform design is not neutral, and that algorithms and ad models are engineered choices shaped by the values of the companies that own them. You can articulate why Malaysia's current approach (post takedowns, content moderation, licensing) misses the structural problem. You feel the drive to push for real transparency and accountability from platforms and regulators.
Engagement signals. Watch-time optimisation. Implicit feedback loops. The actual machinery that decides what you see, and what gets denied visibility.
Why the cheapest content to serve is the most extreme. Why outrage scales better than nuance. Why this is a feature of the model, not a bug in moderation.
Why takedowns and user prosecution miss the point. Why "register with us" is not the same as "show us how the system works." Why Malaysia's framework leaves the engine untouched.
What researcher access provisions actually unlock. What systemic risk audits do. What Malaysia could borrow, and where the gaps are when you try to.
TikTok's aggregated transparency reports. X's "open source algorithm" claim and its limits. Why partial transparency is sometimes worse than none.
What it looks like to engineer for social cohesion and society's well-being. Bridging-based ranking. Exposure diversity. Alternative reward signals. The alternatives exist.
The campaign moves when one person passes a card to another. Building our knowledge base is the first step to being a more conscious social media user.