Social chains

The link that belongs to someone else

Marcus had built a morning chain so reliable it felt mechanical. Alarm at 6:15. Feet on the floor. Bathroom. Brush teeth. Kitchen. Start the coffee maker. While the coffee brewed, pack his daughter's lunch — sandwich, fruit, granola bar, water bottle. Set the lunch bag by the front door. Pour coffee. Calendar review for ten minutes. Out the door by 7:00. The chain had been running for six weeks without a single misfire.

Then his daughter turned three, and the chain collapsed.

The problem was the lunch-packing link. For weeks it had been a solo operation — open the fridge, assemble the items, close the bag. But now his daughter wanted to help. She wanted to choose her own fruit. She wanted to place the granola bar in the bag herself. Some mornings she cooperated in ninety seconds. Other mornings she insisted on a banana, changed her mind to an apple, dropped the apple, cried because the apple had a bruise. The lunch-packing link, which had been a predictable four-minute operation, became a variable ranging from two minutes to fifteen. A twelve-minute lunch negotiation meant the calendar review got skipped, the departure time slipped, and Marcus arrived at work already behind.

His first instinct was to remove the social element — pack the lunch the night before, while his daughter was asleep. This worked mechanically but created a different problem: his daughter's morning became a source of conflict ("I wanted to pick my fruit!") and his relationship absorbed a cost the chain was supposed to prevent. He had optimized the chain by eliminating the human from it, which is the wrong solution when the human matters.

What Marcus needed was a different kind of link. Every link he had designed so far assumed a solo operator executing a scripted action with a predictable outcome. The lunch-packing link involved another person whose behavior he could influence but not control. It was a social link, and social links require a fundamentally different architecture than solo links.

The irreducible unpredictability of other people

Every chain discussed in this phase — morning chains (Morning chains), work startup chains (Work startup chains), shutdown chains (Shutdown chains), exercise chains (Exercise chains) — has implicitly assumed that you are the only actor. You set the alarm. You open the laptop. You put on the running shoes. When every link depends solely on your own behavior, the chain's reliability is a function of your design quality and your practice. Get both right and the chain runs.

Social links break this assumption. A social link is any link in a chain where the completion conditions depend, even partially, on another person's behavior. Making coffee is a solo link. Asking your partner whether they want coffee and waiting for an answer is a social link. Reviewing your task list is a solo link. Having a standup meeting with your team is a social link. Solo links have a single source of variance — you — while social links have at least two. You can script your own behavior to a precise sequence with precise timing. You cannot script another person's behavior at all. You can influence it through requests, agreements, and norms. You can predict it based on past patterns. But you cannot guarantee it, and any chain link that requires a guaranteed outcome from another autonomous human is a link designed to fail.

This is not a flaw in other people. It is a structural property of social interaction. Your three-year-old is not sabotaging your morning routine when she takes seven minutes to choose a piece of fruit. She is being a three-year-old — a person with her own developmental needs, her own timeline. Your colleague who arrives ten minutes late to the standup is not disrespecting your chain. He is navigating his own chain, which may have hit a snag you know nothing about. Designing chains that treat other people's behavior as controllable variables is not just impractical. It is a misunderstanding of what social interaction is.

From scripts to coordination games

The theoretical framework that best captures the dynamics of social chains comes from game theory — specifically, coordination games. In a coordination game, two or more players benefit from aligning their behavior, but alignment is not automatic. It requires communication, convention, or adaptive strategy. Thomas Schelling, in The Strategy of Conflict (1960), introduced focal points — default solutions that players converge on without explicit negotiation. Two people who need to meet in New York City without communicating tend to choose Grand Central Station at noon, not because it is the only option but because both can independently predict the other will choose it.

Social chains operate within coordination game dynamics. When you and your running partner agree to meet at 6:00 AM at the park, you have established a focal point. But your morning chain and your partner's morning chain are both running simultaneously, each with their own links, transitions, and potential breakdowns. When both chains fire correctly, coordination succeeds. When one chain stalls — their toddler woke up early, their car would not start — the coordination fails, and your chain stalls at the social link because the other player did not arrive.

The solution from game theory is not to demand perfect coordination. It is to design strategies that are robust to coordination failures. In behavioral chaining terms, this means designing social links with branching options (Branching chains established the architecture for this) and wider time bounds than solo links. Your solo links can be precise: "four minutes to pack the lunch." Your social links must be ranges: "between two and ten minutes for the lunch-packing interaction, depending on her mood."

Nicholas Christakis and James Fowler, whose work on social contagion was covered in Social habits, demonstrated that behavioral patterns propagate through social networks — but they propagate imperfectly (Christakis & Fowler, 2009). A friend's exercise habit increases your probability of exercising, but it does not guarantee it. The same principle applies within a single chain. A social link increases the probability that an interaction will unfold helpfully, but it cannot guarantee it. Designing for probability rather than certainty is the fundamental shift required when chains become social.

The controllable-core principle

The practical solution to social unpredictability is a design principle you can apply to every social link in every chain: define the controllable core, then flex on everything else.

The controllable core is the portion of the social link that depends entirely on your behavior — the part you can execute regardless of what the other person does. In Marcus's morning chain, the controllable core of the lunch-packing link is: open the fridge, get the lunch bag out, have the sandwich prepared, place the water bottle. These actions happen no matter what his daughter does. The flexible portion is her participation: choosing the fruit, placing the granola bar. That part might take ninety seconds or eight minutes, and the chain needs to accommodate the range without breaking downstream.

Barbara Fiese and colleagues, in their 2002 review of fifty years of research on family routines and rituals, documented exactly this architecture in functional families. Families with robust daily routines achieved robustness not through rigid scripting but through what Fiese called "flexible stability" — a predictable structure that accommodated multiple family members with different needs, ages, and temperaments (Fiese et al., 2002). The parents controlled the structure. The children varied within it. The routine held because it was designed to absorb variation rather than being disrupted by it.

Applying the controllable-core principle requires three design decisions. First, identify what you will do regardless of the other person's response — your anchor within the social link. Second, define the range of acceptable outcomes. Not one specific outcome, but a set: "She chooses a fruit" is the ideal, "she refuses to choose and I pick one" is acceptable, "she has a meltdown and I skip the fruit today" is the minimum viable outcome. Third, define the time boundary. The social link gets a window — not a fixed duration — and the chain proceeds to the next link when either the interaction completes or the window closes, whichever comes first.

This is the critical insight: you do not design the social link to produce a specific result. You design it to produce a result within a range, and you design the next link to accept any result within that range as its input.

Designing social chains in practice

Building a chain that includes social links requires a different workflow than building a purely solo chain. The core steps are sequential, and skipping any of them produces a chain that looks complete on paper but stalls in practice the first time another person deviates from the script you unconsciously wrote for them.

The first step is to map your chain and identify which links are social. Go through your existing chain link by link and ask: does this link's completion depend on anyone besides me? Most people undercount their social links on the first pass. "Make breakfast" feels like a solo link until you realize that your partner's presence in the kitchen changes the timing and the conversational demands on your attention. "Drive to work" feels solo until you remember that you drop your child at school on the way, and the drop-off involves a social interaction with uncertain duration.

The second step is to define, for each social link, the range of acceptable outcomes. Most chain designers fail here because they specify only the ideal outcome and treat it as the only outcome. A robust social link defines two or three outcomes across the likely spectrum: ideal, degraded-but-functional, and minimum-viable. For each outcome, you need to know what happens next. This is where the branching architecture from Branching chains becomes essential. Social links almost always require at least a binary branch: the cooperative path and the non-cooperative path.

The third step is to keep your pre-social and post-social links tight. The flexibility of the social link is not an invitation to make the entire chain loose. The links before the social interaction should be as precisely scripted as any solo chain. The links after should also be tight, designed to accept any of the defined outcomes from the social link and proceed without deliberation. Only the social link itself flexes. The chain's overall structure remains disciplined.

The fourth step is to decide whether the chain should be shared. Some social chains work best when only you know the chain exists — your morning routine includes a brief interaction with your partner, but the chain is yours. Other social chains work best when both people understand and participate in the design. A family bedtime routine functions better when both parents and the child understand the sequence: bath, pajamas, teeth, two books, lights out. When the chain is shared, social links become coordination points rather than unpredictable encounters. Both parties know what comes next, and the chain becomes a collaborative structure rather than one person's private sequence that another person inadvertently disrupts. Fiese's research strongly supports this approach: family routines that were explicitly communicated and collaboratively maintained showed greater stability over time than routines imposed by one member and passively followed by others (Fiese et al., 2002).

The asymmetry of social links

There is a subtlety worth explicit attention: social links are asymmetric. When your running partner shows up three minutes late, your experience is that the chain stalled — downstream links are shifting, your timeline is compressing. Their experience is that they arrived at the park and said good morning. They do not know about your chain. The three-minute delay that feels catastrophic inside your architecture is, from their perspective, trivial.

This asymmetry creates an emotional risk. Chain-aware people tend to feel frustrated not just at the delay but at the other person's apparent indifference to it. But the other person is not being indifferent. They are simply not operating inside the same framework. You can reduce this friction by building time buffers around social links and by sharing the chain so the other person understands the downstream consequences they cannot see. But you cannot eliminate the asymmetry entirely. Other people are not links in your chain. They are autonomous agents. A well-designed social chain respects this by treating the other person's behavior as weather — something you prepare for and adapt to, not something you control.

The Third Brain

An AI assistant can be remarkably useful for designing social chain variants because it can hold the perspectives of multiple participants simultaneously. Describe your chain, identify the social links, and ask the AI to generate response protocols for three scenarios: the cooperative case, the partially cooperative case, and the non-cooperative case. For each scenario, have the AI specify what you do — not what the other person should do. This keeps the design anchored to the controllable core rather than drifting into fantasies about how other people ought to behave.

The AI can also help you identify hidden social links. Describe your chain in full and ask: "Which of these links implicitly depends on another person's behavior?" You will often discover dependencies you had not recognized. "Open the bathroom door" is a solo link until the AI points out that your partner may be in the bathroom at that time, turning a solo link into a social one that requires a branching path (wait, use the other bathroom, or shift the link's position in the chain).

Finally, the AI can serve as a rehearsal partner for the communication step. If you decide to share a chain with a partner, child, or colleague, the way you present it matters. "I have a rigid morning schedule and I need you to comply" produces defensiveness. "I have been thinking about how our mornings work together, and I want to design something for both of us" produces collaboration. Ask the AI to help you frame the conversation as co-design rather than compliance. The strongest social chains are the ones both participants feel ownership over.

Chains do not maintain themselves

Social chains carry a particular maintenance burden because they involve at least two people whose lives change over time. The running partner who was reliable at 6:00 AM gets a new job with an earlier start. The child who cooperated with the bedtime chain at age four resists it at age six. The colleague whose standup anchored your work startup chain transfers to a different team. Every social link has a shelf life determined not by your consistency but by the other person's changing circumstances.

This means social chains require more frequent review than solo chains. You cannot set them and forget them. In Chain maintenance, you will learn the broader discipline of chain maintenance — periodic review and adjustment of all your chains, social and solo alike. The principles covered there will apply with special force to social chains, where the variables include not just your own habits and energy levels but the evolving lives of the people your chains depend on.

---

Sources:

1. Schelling, T. C. (1960). The Strategy of Conflict. Harvard University Press.
2. Christakis, N. A., & Fowler, J. H. (2009). Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives. Little, Brown and Company.
3. Fiese, B. H., Tomcho, T. J., Douglas, M., Josephs, K., Poltrock, S., & Baker, T. (2002). "A review of 50 years of research on naturally occurring family routines and rituals: Cause for celebration?" Journal of Family Psychology, 16(4), 381-390.
4. Gollwitzer, P. M. (1999). "Implementation intentions: Strong effects of simple plans." American Psychologist, 54(7), 493-503.
5. Wood, W., & Neal, D. T. (2007). "A new look at habits and the habit-goal interface." Psychological Review, 114(4), 843-863.