The Accretion Frame

The accretion frame is the design theory behind a pull engine. A pull engine is a practice that produces its own return visits. A long practice runs on one of two things. It runs on push, which means forcing each session against not wanting to go. Or it runs on pull, which means the structure of the practice makes the next visit happen on its own. Wanting to return comes from the structure.

Pull holds when three conditions are present at the same time. The practice grows a persistent artifact. It leaves a next piece visibly unfinished. A missed day costs nothing. When all three are present, the return visit does not have to be forced.

Push has its own set of tools: defusion, willpower budgeting, and environment design. These are the tools of Self-Regulation and Self-Management. The accretion frame is designed so that these tools do not have to be used every day.

Core takeaways

• A practice runs on push or on pull. Push means forcing each session. Pull means the structure of the practice brings the next visit on its own.
• Pull needs three things at once: an artifact that persists and grows, a next piece left visibly unfinished, and a missed day that costs nothing.
• A streak or a closed ring counts days in a row against a daily requirement. Its value depends on staying unbroken, and a rest day counts as a failure.
• Tracking an enjoyable activity raises how much gets done and lowers enjoyment and later participation. So the number belongs on the collection, not on the session.
• How a session feels while it is happening predicts how much activity is still going on six months and a year later.
• The design removes the effort of showing up. The effort of the work inside the session stays.

What a streak or a closed ring does

A streak or a closed ring is a maintenance metric. It counts days in a row against a daily requirement. The number has value only while it is unbroken. So the larger the number grows, the more there is to lose by breaking it. Once the chain exists, the day’s decision changes. It stops being a decision to add a piece and becomes a decision to protect something already owned.

Keeping the chain feels good at first and then feels good less and less. Breaking it keeps hurting. Over a long enough run, the metric does more harm than good. Anything short of the daily requirement counts as nothing. After one miss, a second miss costs almost nothing extra. The same collapse has been documented when a diet rule is broken, and it is a fair description of what a streak does. Products ship freeze buttons because a broken chain is a known reason people quit.

What tracking costs

Tracking has a cost. Six experiments on enjoyable activities found that tracking raised how much got done, lowered how much it was enjoyed, and lowered how much of it happened later. That damage applies to leisure. Goal-directed output is mostly unaffected.

A daily requirement also uses the wrong time scale for a body. Capacity builds across weeks, and it needs rest to build. Scoring a rest day as a failure scores correct training as a failure.

The parts that make pull

Remove each part in turn and check whether the pull is still there. Remove the thing that accumulates, and the wish to return stops. Remove surprise, so that every session is the same, and the wish to return stops. Remove the quota, and nothing changes. Remove the other people, and starts get slower, but the pull stays. Two parts do the work: a thing that only grows, and a next addition whose result is not known in advance.

In practice those two parts are one property. The next addition is visible and its outcome is open. A ring shows a gap that has closed. A street map with unwalked blocks shows a gap that is still open. Regret on a skipped day, felt without any app, is the person noticing that the artifact did not grow that day.

What the maintenance research says

The research on keeping up exercise supports pull, though it does not describe this three-part design. Motives of enjoyment, competence, and belonging predict staying. Motives of health, weight, and looks predict a start that later stops. The feeling inside one session predicts how much activity is still happening half a year and a year later. That feeling matters as much as the artifact does. An eight-week trial set intensity by pleasure instead of by fitness numbers. It got 77 percent more attendance: 14.35 sessions against 8.13, with forty-six people retained. It was a small trial. The thing that changed was how the hour felt while it was going on.

The idea of willpower as a muscle that runs out has mostly failed to replicate. It is a rejected theory and does not need to be answered.

Climbing, skating, dance, and martial arts already have this structure. In each of them, skill is a tree that keeps growing and never runs out, and the scene around the activity holds the structure. The word these activities use is autotelic: the activity is its own reason.

The time scales a body keeps

A body changes on more than one time scale. The slow changes survive a gap. A motor skill lasts years. Tendon and connective tissue rebuild over months. Bone turns over in a cycle of three to four months. Muscle that has been trained before comes back much faster than muscle built for the first time.

The fast changes come and go from week to week. Stored carbohydrate, plasma volume, mitochondrial enzymes, and the mood lift after a session are real and useful, and they are gone if they are not restocked.

One skipped day cannot remove anything structural. A skipped week often leaves capacity level or higher, because tiredness goes away faster than capacity does. Mood, insulin response, and metabolic sharpness need a rhythm well below daily: hard breathing at some point inside a loose window of two or three days.

There is a check that can be applied to any practice. After a month completely off, does the number still mean anything? Unlocked skills, total distance, dated bests, and named routes still mean something. Day-chains, closed rings, and calories burned do not.

Seven design rules

There are seven rules. Each one prevents one specific failure.

Rule	What it prevents
Append-only structures	A pile that can shrink, and with it a sense of going backwards
No denominators	A hundred-percent target that turns absence into falling behind
Check after, never before	The log becoming an audit that decides whether the day is permitted
Body untracked	Daily body numbers bringing a maintenance metric back, built on the noisiest data a body gives
No numbers on the enjoyed part	Measurement placed on the experience instead of on the collection
Rest is correct	A rest day scored as a failure
Reward the return	Punishing the comeback, when the research says to pay for it

Tracking lowers enjoyment of leisure and mostly leaves goal-directed work alone. So the number belongs on the collection, not on the hour itself. In a study of more than sixty thousand gym-goers, the arm that worked best paid a few cents for showing up again after a skip. The payment was about nine cents, and it produced a 27 percent rise in visits. One skipped day does not undo the automaticity already built. The line often quoted as a sixty-six-day clock is the wrong line from the same paper.

Small starts, pairings, and standing times

The start of a session stays small: sets below a maximum, scraps of movement, a first action too small to need a decision. Pairing the hour with something already wanted raised gym attendance 51 percent in its field trial. That figure is the lift at the start. It faded. It is not a permanent 51 percent.

Showing up on schedule is mostly a property of the surroundings. People who look disciplined are usually people whose defaults already make the decision. About 45 percent of what a day contains happens again in the same setting. A standing time, place, class, or partner survives a gap. One miss costs nothing, and coming back is the default.

What the frame does not fix

A day-chain does give cheap structure, and it holds while the activity still fits who the person is. The collapse after a break is also real. Both are true. The chain does not have to be thrown away.

Pleasure during movement drops once conversation becomes hard. That point is the ventilatory threshold, and the talk-test is the field version of it. Intensity that someone else imposed is what gets abandoned. Nothing that accumulates will save a session that feels bad while it is happening. The person doing the session chooses the speed. The session stops while it still feels good, because the memory of that feeling is what predicts the next session.

Activities that run on pull still have a hard stretch. The plateau lasts. The struggle comes before the flow. The design makes it easy to arrive at the session. The work inside the session stays as hard as before.

Two checks test the design. Three weeks in, a skipped day should produce a specific picture of an artifact that did not grow. If it produces nothing, the artifact is not doing anything. A planned week off sorts the rest. “The pile paused” means the design is holding. “Behind schedule” means a rate has crept in. Remove the denominator.

Where push tools still belong

Push tools stay for domains where nothing accumulates, and for cold starts inside domains that have been converted to pull. DFUZ is a one-session move that unhooks the person from the feeling of not wanting to start, so a stalled practice can restart. It is used only as a reserve, not every day. The cost of the frame is design work. The first live conversion is Movement as Accretion. The Treadmill Library is the listening arm of that conversion. The working name for a movement practice that never uses the motivation machinery is motivationless movement. It is a working name only.

Wanting to return still comes from structure. The structure has to include how the session feels, and it has to leave rest unpunished. The design removes the effort of showing up. The effort of doing the work stays.

How to practice this

1. Three weeks into a practice, skip a day. Notice whether a specific picture of the artifact not growing comes to mind. If nothing comes to mind, the artifact is not doing anything.
2. Take a planned week off. If you say the pile paused, the design is holding. If you say you are behind schedule, a rate has crept in.
3. Ask whether your number would still mean anything after a month completely off. Unlocked skills, total distance, dated bests, and named routes keep their meaning. Day-chains, closed rings, and calories burned do not.
4. Check the log after the session, never before. Notice whether the log had been deciding whether the day was permitted.
5. Choose your own speed. Stop while the session still feels good. Notice the point where talking becomes hard, because pleasure drops past it.
6. Set a standing time, place, class, or partner. After one miss, notice whether coming back happens by default.

Related pages

• Self-Regulation: push-engine tools this frame is designed not to use daily
• Self-Management: the other half of the push toolkit
• DFUZ: reserve move for a cold start, never the daily motor
• Movement as Accretion: the first live conversion of this frame
• The Treadmill Library: the listening arm of that conversion

1 more related pages

• Fitness Mindsets: rules for keeping training and eating going for years, such as saying no and fixed time slots. It tracks the body for a few weeks, while this frame keeps the body untracked.

Open questions

Does this named three-part artifact design outpredict in-session affect alone?

Sources

Etkin, J. (2016). “The Hidden Cost of Personal Quantification.” Journal of Consumer Research 42(6), 967–984. Six experiments: measuring an enjoyable activity raised output and lowered enjoyment and later engagement. Boundary: leisure, not goal-directed output.

Cochran, W., & Tesser, A. (1996). The “what the hell” effect. Herman, C. P., & Mack, D. (1975). Counterregulatory eating. Everything below the bar reads as zero; once missed, the cost of continued lapse collapses.

Ryan, R. M., Frederick, C. M., Lepes, D., Rubio, N., & Sheldon, K. M. (1997). International Journal of Sport Psychology. Teixeira, P. J., Carraça, E. V., Markland, D., Silva, M. N., & Ryan, R. M. (2012). International Journal of Behavioral Nutrition and Physical Activity. Systematic review, 66 studies. Enjoyment, competence, and connection predict maintenance; appearance and weight motives predict starting then quitting.

Williams, D. M., et al. (2008). Affective response during exercise predicts physical activity at 6 and 12 months. Rhodes, R. E., & Kates, A. (2015). Review confirming the direction.

Teixeira, D. S., et al. (2024). International Journal of Behavioral Nutrition and Physical Activity 21:85. Preregistered pragmatic trial. Intensity prescribed for pleasure vs fitness targets: 14.35 vs 8.13 sessions over eight-week follow-up (77%). N=46 retained.

Hagger, M. S., et al. (2016). Perspectives on Psychological Science. 23 labs, N=2,141, ego-depletion effect not significant. Vohs, K. D., et al. (2021). Multilab failure. Willpower-as-muscle as a discarded rival.

Milkman, K. L., et al. (2021). Nature 592, 390–394. N>60,000. Top arm: a small bonus for returning after a missed scheduled workout, about $0.09, 27% boost in gym visits.

Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). European Journal of Social Psychology. Missing one day did not materially affect the automaticity trajectory.

Milkman, K. L., Minson, J. A., & Volpp, K. G. (2014). Management Science. Temptation bundling raised gym attendance 51% initially. The effect decayed (Kirgios et al. 2020).

Wood, W., & Neal, D. T. (2007); Wood, W., & Rünger, D. (2016); Neal, D. T., Wood, W., & Quinn, J. M. (2006). About 45% of daily behavior repeated in the same context.

Ekkekakis, P., Hall, E. E., & Petruzzello, S. J. (2004, 2008); Ekkekakis, P., Parfitt, G., & Petruzzello, S. J. (2011). Dual-mode theory: affect turns negative near the ventilatory threshold. Talk-test as the field proxy.

Issurin, V. B. (2010); Mujika, I., & Padilla, S. (2000); Banister impulse-response / fitness-fatigue. A missed day cannot cost anything structural; fatigue’s time-constant is shorter than fitness. Motor-skill retention, tendon collagen turnover, bone remodeling (~3–4 months), and muscle memory: Park & Shea; Magnusson / Heinemeier & Kjaer; Frost; Gundersen 2016; Seaborne et al. 2018.