Your Attention Belongs to Someone

If you don't have an algorithm, you're being run by one

Most people think of an algorithm as an invisible machine that learns their preferences. That image is misleading. An algorithm is simply a set of rules for deciding what gets your attention next. Every platform that surfaces content for you is running one. The question is not whether you have a relationship with an algorithm. It is whether that algorithm is really yours.

The answer, for most people, is no.

What the algorithm is optimised for

The platforms that mediate modern information consumption — social media feeds, video recommendation engines, news aggregators — optimise for one thing above everything else: continued engagement. Not understanding. Not wellbeing. Not the quality of your decisions. But time on platform.

This is not a conspiracy. It is a business model, implemented at scale, and it works. The most reliable way to keep someone watching turns out to be triggering emotion — particularly the emotions clustered around social comparison: outrage at what others are doing, desire for what others have, anxiety about being left behind. These aren't bugs; they're the predictable output of systems trained on human behaviour at enormous scale.

The result is that the status cycle — the loop in which status-seeking drives consumption, consumption drives resource demand, demand drives scarcity, scarcity drives competition — is not only a feature of human psychology. It is amplified by infrastructure that profits from keeping it running. The platforms didn't design this outcome deliberately; they trained for engagement and this is what engagement-maximisation produces.

Attention is upstream of desire

Most conversations about changing behaviour focus on the desire end of the cycle: what people want, what they value, what they believe is worth pursuing. Rutger Bregman's argument in Moral Ambition is an example of this — the case that we should redirect social recognition toward people doing work the actually matters, making moral contribution the currency that status-seeking runs on. That is a powerful intervention, and a necessary one.

But desire begins with attention. Before you want something, you see it. Before you believe something matters, you encounter it repeatedly enough to take it seriously. The algorithm that shapes your feed shapes the inputs to your wanting before conscious evaluation has a chance to operate. Changing what you desire without changing what you attend to is like trying to eat less while someone refills your plate.

This is not a new observation. Media literacy, curated reading lists, intentional information diets — all of these are attempts to manage attention deliberately. What has changed is the scale and sophistication of the systems working in the other direction. The platforms have vast data on what keeps each individual watching. The individual has their own judgment, intermittently applied. The asymmetry matters, but it is not insurmountable.

The personal algorithm

The most tractable response is not primarily regulatory — waiting for platforms to open their recommendation systems, or for governments to mandate transparency. Those efforts matter, but they are slow and contested. The more immediate option is architectural: build a personal layer that sits between the raw feed and your attention.

A personal algorithm is an explicit set of choices about what you want to notice: which sources you trust, which topics deserve your time, what quality of argument you're willing to engage with, what you want to see less of. At its simplest, it is a curated reading list and a habit of using it. At its most developed, it is a system that learns from your responses, weights your stated values, and updates as your understanding evolves.

The principle is the same at both levels of complexity. Attention directed by intention differs in kind from attention captured by engagement optimisation. The gap between them is where a significant portion of the status cycle operates — and where the most tractable individual intervention lives.

The foundation for a personal algorithm begins with four questions: Who am I? What do I believe is important? What am I actually capable of? What do I want to do? These aren't abstract philosophical exercises — they are the inputs that determine which sources are worth your time, which topics are load-bearing for your thinking, and which stories are signal versus noise for your actual life. Without answers to them, a curated feed is just aesthetic preference. With them, it becomes an expression of values applied to attention.

The overlay model

Personal algorithms are unlikely to run natively on the platforms that currently control most people's information environment — at least not initially. The platforms have limited incentive to cede control over what users see. But they don't need to. A personal algorithm can run as an overlay: a layer of intentionality that sits on top of existing infrastructure, shaping what you ask for and how you interpret what you receive, without requiring the platform's cooperation.

This model already works in adjacent domains. Email clients filter and prioritise without the sender's involvement. RSS readers aggregate across sources on the reader's terms. The memory systems people now build on top of AI assistants — explicitly encoding context, values, and preferences that the underlying model doesn't carry by default — are another instance of the same principle. You don't rebuild the infrastructure. You add a layer that represents you rather than the platform's interests.

The friction of building that layer is the real barrier. Most people do not have the time, inclination, or technical confidence to construct an explicit information diet, let alone a feedback loop that updates it. The tools that make this tractable — that lower the activation energy between "I want to manage my attention differently" and "I have a working system for doing that" — are what's missing. They don't need to be complex. The simplest version is a structured list of sources, topics, and exclusions, applied consistently. That alone is a significant departure from the default.

What this changes — and what it doesn't

A personal algorithm does not solve the attention problem. It is a partial, individual response to a structural condition. The platforms will continue to optimise for engagement. The asymmetry between commercial systems and individual users will persist. The collective action problems that make systemic change difficult are not resolved by any individual building a better feed.

What it does change is the relationship between your values and your attention. Most people, if asked, could articulate something about what they think matters: what kind of work is worth doing, what kind of world is worth building, what they want to understand better. Most people's information environment has no relationship to those articulated values. The algorithm running their feed has never been asked. Building a personal algorithm is the act of asking — and then making the answer consequential.

That is a modest thing at the individual level. It is not a modest thing in aggregate. The status cycle runs partly on attention that has not been claimed. Claiming it, intentionally, is both a tractable entry point and a model for what more systematic change might eventually look like.

This essay is part of a series on tools, cycles, and the problems worth working on.

Written with AI. Version 1 — March 2026.