Examining dependence in digital systems requires us to drop our focus on individual blame, because the usual arguments of “users choose,” “people can leave,” “nobody is forcing them,” and so on sounds plausible only when we forget the conditions that produce dependences, as though choice were something exercised in a vacuum rather than inside architectures calibrated around fatigue, habit, loneliness, scarcity, and the managed erosion of attention. Agency is real, but it is exercised through bodies that get tired, minds that fragment, incomes that tighten, schedules that collapse, and social worlds already saturated by metrics and prompts. Why do we ask people to resist systems designed to wear resistance down? And what would responsible design look like in an economy that rewards dependence?

Dependence is rarely produced through one dramatic mechanic, but through soft pressures that accumulate without making the coercion obvious. For example, a streak that converts absence into failure, a reward loop that never quite finishes, a timer that turns waiting into a pain point, a shop that obscures prices behind currencies built to obscure actual spending, or a progression curve that begins with momentum but becomes constant maintenance. None of this needs to be spectacular to be effective. In fact the opposite is true. The most reliable systems of extraction are often the ones that feel ordinary, becoming part of how we wake, check, respond, optimise, and return. The harm appears diffuse only because it has been distributed so efficiently across time.

So any serious solution has to begin with the structure itself. We need to build engagement systems that do not require depletion in order to function. That means treating attention as something with a social cost, not an infinitely harvestable resource. If a product can only survive by turning interruption into routine, anxiety into retention, and incompletion into revenue, then the design problem is economic. We are dealing with a business model whose success depends on converting human limits into recurring income. No amount of ethical language can neutralise that process. The work begins when we accept that humane outcomes require material constraints on what the system is allowed to optimise for.

One of those constraints asserts that monetisation should follow value delivered, not friction imposed. When systems profit by slowing progress and then selling relief, by withholding convenience and then pricing its return, they become structurally incapable of respecting the user because the user’s discomfort is no longer a side effect but a revenue source. A healthier design relation emerges when payment attaches to substantial content, durable ownership, or clearly bounded service rather than to acceleration through bottlenecks. This does not solve everything, but it changes the direction of incentive. It becomes possible to make something better by making it richer, not by making it harder to put down.

The same principle applies to time. A system that respects human life must know how to end, or at least how to pause without punishment. We should design for completion, breathable intervals, and moments where the interface acknowledges that enough has happened for now. This sounds modest until we recognise how much of contemporary design is built around abolishing exactly that feeling. Endless feeds, rolling objectives, rotating events, and perpetual optimisation all serve the same function: they prevent closure because closure interrupts capture. To design against dependence is therefore to make stopping possible and safe. Not hidden three layers deep behind the next prompt, not technically available while socially penalised, but openly present as part of the intended use of the system. A product that cannot tolerate our absence has already told us what kind of relationship it wants.

Clarity matters just as much. Abstract currencies, fragmented subscriptions, probabilistic rewards, and obscured conversion rates do more than confuse; they weaken the user’s ability to map action to consequence. That weakening is politically important because it shifts power toward the operator of the system, the actor with complete visibility over pricing, timing, and behavioural response. Transparent design redistributes some of that power back outward. Direct prices, plain language, visible tradeoffs, and honest probabilities do not eliminate exploitation by themselves, but they reduce the informational asymmetry through which exploitation usually works.

We also need to reject the fantasy that every system should scale infinitely upward in intensity. Infinite ramps are mechanisms for keeping desire in motion long after satisfaction would otherwise have arrived. A healthier progression model knows how to plateau, how to deliver competence without converting mastery into endless maintenance, how to let people feel finished without framing that feeling as churn risk. There is something deeply political in allowing an experience to be complete. Capital prefers subjects who remain slightly behind, slightly underrewarded, slightly unfinished, because unfinished people are easier to monetise. Designing for enoughness is therefore not merely a user-experience decision. It is a refusal of the economic logic that treats all settled satisfaction as a lost opportunity.

Then there’s unequal vulnerability: children, neurodivergent people, the overworked, the lonely, the financially stressed, those already living in fragmented conditions of attention and care, encounter engagement systems from where they actually are, which means harms compound along existing lines of inequality. Good design, if it deserves the word, must be built around that asymmetry rather than around an imaginary average user with stable resources and unlimited resistance. Protective defaults, slower loops, clearer boundaries, fewer manipulative prompts, and an aversion to obligation-based retention should be treated as baseline design commitments, because a system that is safe only for the well-resourced is not safe in any socially meaningful sense.

Of course this is where the horizon of reform becomes visible. Many of these choices are entirely achievable in technical terms and deeply inconvenient in financial ones. Venture-scale expectations actively narrow the field of what can be chosen. When firms are valued on growth curves that presume continuous expansion of captured time, even modest restraint starts to look like underperformance. So a solutions-focused approach has to be honest about scale and ownership. Some harms can be reduced through better design practice within existing institutions, but if the revenue model remains organised around extracting more value from attention year after year, humane design will remain fragile, partial, and permanently under pressure. The solution is not just better principles inside the same machine. It is different tolerances for profit, different governance structures, different definitions of success.

That may sound too large, but it is actually the most practical thing we can say. We do not need to wait for a total transformation before acting. We can build systems that charge plainly, end gracefully, refuse artificial scarcity, and decline to turn psychological pressure into product strategy. We can normalise the idea that a digital experience should leave people more intact than it found them. We can create teams where the measure of success includes what the system does not take. We can make it easier for designers and engineers to argue against manipulative mechanics by naming the economic conflict directly rather than hiding it behind vague appeals to wellbeing. Precision is useful here. Once we identify the incentive, we can identify the point of intervention.

And what comes after dependence is not emptiness, but a different kind of design problem. If we refuse systems that survive by manufacturing need, then we have to become serious about making forms of difficulty that do not degrade the people who meet them. That is part of what makes An Interesting Solution to an Interesting Problem feel so resonant here: it turns toward the question of what people create when extraction is no longer the only engine available, and suggests that the real challenge is not how to keep subjects trapped inside artificial scarcity, but how to build worlds with enough freedom, enough texture, and enough genuine stakes that agency can become interesting again.

So the question is no longer how we keep people engaged forever, because forever has always been the language of extraction, the dream of a system that never has to release its hold. The better question is what forms of engagement remain worthy once we refuse dependence as a design objective. What do we build when our goal is not to capture more life, but to leave more of it available for living.