Self-improvement usually begins with something visible. Sleep is poor, so we work on our sleep. Exercise has disappeared, so we design a routine. Food has become erratic, so we clean up the kitchen. Attention is fractured, so we remove notifications.

There is nothing wrong with starting there. A specific behavior is easier to change than an entire life, and good interventions often do begin with one concrete action.

The trouble starts when a behavior is treated as if it exists on its own.

A better bedtime has to compete with work that never quite closes. A better diet has to survive the evening created by the rest of the day. A boundary has to survive the financial and social consequences of holding it. A workout plan has to fit inside sleep, travel, family obligations, energy, and the calendar rather than beside them.

When the new behavior repeatedly loses that competition, the explanation usually contracts around the person: not disciplined enough, not motivated enough, not serious enough.

Sometimes that diagnosis is fair. Often the unit of analysis is simply too small.

The visible behavior is downstream

Habit research gives us a useful piece of this. Habits are not stored as isolated intentions. They become associated with the contexts in which they are repeatedly performed. Wendy Wood and David Neal described habits as learned links between responses and recurring features of the performance context. More recent work found that people trying to build habits showed greater automaticity and better goal attainment when the context of repetition was more stable.

That does not mean we need to turn life into a controlled laboratory. It means the setting participates in the behavior.

Consider a person trying to stop scrolling at night. The obvious intervention is to restrict the phone. But the phone may be the last available source of stimulation after a day with no pleasure, the easiest way to avoid the unfinished feeling of work, or the cheapest form of downshift available to a tired nervous system. Removing the device may help. But it may also leave the demand that made the device useful untouched.

The same logic appears in food, work, exercise, and relationships. What looks like one bad choice is often the end of a longer sequence: poor sleep narrows capacity, narrower capacity changes decisions, those decisions change the day, and the altered day makes the next repetition more likely.

By the time the visible behavior occurs, much of the negotiation has already happened.

Knowledge is often not the scarce resource

This is an unusual moment in the history of advice. For many common health and performance problems, basic information is abundant. Most people who read material like this already know that sleep matters, movement matters, food quality matters, chronic stress has consequences, and phones can fragment attention.

Yet knowing and doing remain stubbornly distant.

That gap is frequently interpreted as a failure of motivation. But behavior change research has increasingly had to account for social and economic conditions because interventions do not land in neutral lives. A review led by Carmela Alcántara noted that social determinants can shape the effectiveness of behavior-change interventions, while also pointing out how unevenly those conditions have been studied.

The practical implication is less grand than the term “social determinants” makes it sound.

A worker can understand the cost of late-night availability and still depend on the job. A parent can value exercise and still have no reliable period of the day that belongs to them. A founder can know that constant responsiveness is unsustainable and still be carrying payroll. A caregiver can understand the need for rest and remain the person everyone else depends on.

The information can be correct while the conditions remain unchanged.

A new habit has to enter an old architecture

This is where many self-improvement systems become unintentionally moralizing. They isolate the final action and then evaluate the person by whether it occurred.

But the environment has been rehearsing the older behavior for much longer than the new intention has existed.

If work expands into the evening, the calendar has one pattern. If the kitchen contains nothing easy at the moment of greatest fatigue, it has another. If every social event is organized around a behavior someone is trying to reduce, the social environment adds its own pressure. If the person has always been rewarded for absorbing excess responsibility, even a sensible boundary can feel like a threat to identity.

None of those conditions removes agency. They tell us where agency has to operate.

Sometimes responsibility means performing the useful behavior despite the difficulties. Sometimes it means changing the environment that keeps making the behavior difficult. Sometimes the harder intervention is not the workout, the meal, or the bedtime. It is the conversation, the staffing decision, the shopping pattern, the financial tradeoff, or the end of the workday.

That is a larger form of responsibility than telling ourselves to try harder.

Change the condition that keeps producing several problems

Once we widen the frame, the problem can actually become simpler.

The aim is not to optimize every variable. That would create a larger self-improvement project and another set of things to fail at.

The useful question is which condition is producing several of the outcomes we keep trying to fix separately.

Sometimes poor sleep is upstream of food, attention, training, and patience. Sometimes work design is upstream of poor sleep. Sometimes the phone is less a cause than a compensation mechanism for a day with no closure. Sometimes inconsistency with exercise is what happens when movement has been assigned to the part of the day that receives whatever capacity is left.

The intervention changes when the sequence becomes visible.

This is the layer I have been trying to make legible in the Human OS work. Not because every problem is secretly the same problem, and not because every difficulty needs a systems framework. Some problems really are local. Some habits really do improve through repetition and discipline.

But when the same outcome keeps returning despite good information and sincere effort, it is worth asking what keeps recreating the conditions for it.

That question has followed me through writing about food, cities, attention, work, technology, identity, and recovery. The subject changes. The recurring discrepancy is that we often judge the person at the point where the larger sequence finally becomes visible.

The Human OS Manual launches this October. I wrote it for people who already know a great deal about what they should be doing and have discovered that more advice is not the same thing as a more coherent life.

If this helped you name something you have been feeling but had not quite put into words, please share it with someone who might need the same language.

The Human OS Manual is now available for pre-order. If this work resonates, you can reserve your copy and follow the ideas as they continue to develop.

Feel free to leave a comment. I read and respond to every thoughtful note.

Notes and sources

Wendy Wood and David T. Neal, “A New Look at Habits and the Habit-Goal Interface,” Psychological Review (2007).
Marco Stojanovic, Axel Grund, and Stefan Fries, “Context Stability in Habit Building Increases Automaticity and Goal Attainment,” Frontiers in Psychology (2022).
Carmela Alcántara et al., “Social Determinants as Moderators of the Effectiveness of Health Behavior Change Interventions: Scientific Gaps and Opportunities,” Health Psychology Review (2020).

Continue through the Human OS

This essay is one expression of the wider architecture. Follow the connections below to move from the observation into the framework, evidence, or book.

Human OS connections

Related evidence

Next step

Explore The Human OS Manual →