Chain documentation

The chain he thought he ran

Marcus had been running his work-startup chain for six months and was proud of it. He was the kind of person who talked about systems at dinner parties. "I've got it down to six steps," he would say. "Park, bag at desk, laptop open, check Slack, review calendar, deep work." He could recite the sequence without thinking, which he took as evidence that the chain was working. Then his executive coach, who had been trained in behavioral analysis, asked him to do something he had never done: write it down. Not recite it. Write it. Every link. Every physical action. Every transition from one behavior to the next.

Marcus sat at his desk with a blank sheet of paper and started writing. Park car. Walk to building. Badge in at the door. He paused. He had never counted badging in as a link because it took two seconds and was completely automatic. But it was a discrete physical action — card out of wallet, tap on sensor, wait for the beep, push door open, card back in wallet. He kept writing. Walk to elevator. Press button. Ride up. Walk to desk. Drop bag. But then he noticed something he had genuinely never registered: he did not go to his desk first. He stopped at the coffee station. Every single morning. He filled his water bottle, poured a cup of coffee, exchanged a few words with whoever was there, and then walked to his desk. This was a three-minute behavior that he had never once included in his mental model of the chain because it did not feel like a "step." It felt like a gap between steps — dead time, transition space, the kind of thing that happens between the real behaviors. But on paper, it was undeniably a link. It was a specific set of physical actions that occurred in a fixed sequence at a fixed point in the chain.

He kept going. Bag at desk. Sit down. Open laptop. While the laptop booted — thirty seconds, sometimes a minute — he picked up his phone and scrolled. Not purposefully. Not checking anything specific. Just filling the wait. Another link he had never counted, never noticed, and certainly never designed. Laptop open. But he did not go to Slack first, despite telling himself for months that Slack came first. He opened email. Every time. The browser defaulted to his inbox, and he would scan subject lines for anything urgent before navigating to Slack. His six-link chain was already at nine links, and he had not yet reached the calendar review.

Two more surprises emerged. "Check Slack" was not a single behavior. It was a decision tree. Which channel first? Did he scan all channels chronologically or jump to the one with the most unread messages? On some mornings he checked the team channel first; on others, the leadership channel. This was not an automatic link — it was an active deliberation disguised as a habit. And "deep work on the highest-priority task" was not a behavior at all. It was an aspiration containing an embedded decision: what counts as highest-priority? Some mornings that decision took thirty seconds. Other mornings it consumed ten minutes of comparison and second-guessing, and by the time he chose, his thirty-minute deep work block had eroded to twenty.

Marcus stared at the paper. His "six-step chain" was actually an eleven-step chain with two decision points, an unacknowledged phone habit, and a coffee detour he had never designed but ran every single day. Nothing was broken. The chain worked well enough. But the documentation revealed a system he had never actually seen, and the gap between the system he was running and the system he believed he was running was the gap where optimization opportunities lived.

Why automatic behavior is invisible

The reason Marcus had never seen his own chain accurately is not a personal failing. It is a structural feature of how automaticity works. Wendy Wood's research on habitual behavior, synthesized in Good Habits, Bad Habits (2019), demonstrates that as behaviors become automatic, they progressively withdraw from conscious awareness. This is the entire point of habit formation — to offload recurring behavioral sequences from the prefrontal cortex to the basal ganglia, freeing executive function for novel problems. But the consequence is that the more automatic a behavior becomes, the less accurately you can self-report it.

Wood and her colleague David Neal conducted studies showing that habitual behaviors persist even when the outcomes they produce change. In one well-known experiment, moviegoers who had a strong habit of eating popcorn at the cinema continued eating stale popcorn they rated as unpleasant, while non-habitual moviegoers stopped eating when the popcorn was stale (Neal, Wood, Wu, & Kurlander, 2011). The habitual eaters were not choosing to eat bad popcorn. They were not even aware they were eating it. The behavior was running beneath the threshold of conscious monitoring, which meant their self-report about what they were doing in the theater would not have included "mindlessly eating terrible popcorn."

This is the fundamental problem that chain documentation solves. Your chains are running, but you cannot see them. You can see the first link because it interfaces with conscious life — you are aware of parking the car or pressing the alarm button. You can see the last link because it produces the outcome you care about — the journal is written, the workout is complete. But the middle links, and especially the transitions between links, have been compressed by the basal ganglia into a chunk that runs without executive oversight. Ann Graybiel's research, which you encountered in Chain anchors, demonstrated this compression directly: as behavioral sequences become habitual, neural activity in the striatum concentrates at the beginning and end of the sequence, with suppressed activity during the middle (Graybiel, 2008). The middle is automated. The middle is invisible. And the middle is where most of your chain's actual structure lives.

Documentation forces precision because it requires you to translate automatic behavior back into explicit, sequential descriptions. You cannot write "then I do my morning routine" on the document. You have to write each action, each transition, each physical movement. The act of writing recruits the prefrontal cortex to re-examine a sequence that the basal ganglia have been running without cortical supervision. This re-examination is uncomfortable because it reveals the gap between your explicit beliefs about your behavior and the actual behavior the habit system is executing.

Making tacit knowledge explicit

The value of documentation extends beyond merely listing behaviors. It converts tacit knowledge — knowledge you have but cannot articulate — into explicit knowledge that can be examined, shared, and improved. Ikujiro Nonaka and Hirotaka Takeuchi formalized this distinction in The Knowledge-Creating Company (1995), arguing that organizational innovation depends on the continuous conversion between tacit and explicit knowledge. Their SECI model describes four modes of knowledge conversion: socialization (tacit to tacit), externalization (tacit to explicit), combination (explicit to explicit), and internalization (explicit to tacit).

Chain documentation is an act of externalization. Your chain exists as tacit procedural knowledge — you know how to do it, but you cannot fully describe it. Writing the chain down converts that tacit procedure into an explicit representation that can be analyzed, debugged, and optimized. Without externalization, your only tool for improving a chain is trial and error: you notice the chain is not working, you try changing something, you see if it gets better. With externalization, you can perform the analysis that trial and error cannot: you can see the entire structure at once, identify the weak joints before they fail, and design targeted interventions rather than random adjustments.

Nonaka and Takeuchi emphasized that externalization is the most difficult and most valuable of the four knowledge conversions. It requires deliberate effort because tacit knowledge resists articulation — it is, by definition, the knowledge you do not know you have. This is why Marcus never noticed the coffee station stop or the phone scroll during laptop boot. These were not things he had decided to conceal from himself. They were genuinely invisible to his self-model because they had been absorbed into the tacit execution of the chain. The documentation process forced them into the explicit domain where they could become objects of analysis rather than invisible features of the landscape.

K. Anders Ericsson's research on deliberate practice reinforces this point from a performance perspective. In his landmark studies of expert performers, Ericsson found that what distinguishes experts from experienced non-experts is not simply the amount of practice, but the quality of representation they maintain of their own performance (Ericsson, Krampe, & Tesch-Romer, 1993). Expert musicians do not just play their pieces repeatedly. They maintain explicit, detailed mental representations of what each passage should sound like, how their fingers should move, where the difficult transitions are. These representations allow them to identify specific errors and target specific weaknesses. Non-experts practice without explicit representations and therefore cannot diagnose their own failures precisely.

Applied to behavioral chains, the lesson is direct. You are an experienced non-expert at your own habits. You have run them thousands of times, but you have never built an explicit representation of their structure. Without that representation, you cannot practice deliberately — you cannot identify which specific link or transition needs work, because you cannot see the links and transitions clearly. Documentation creates the representation that makes deliberate improvement possible.

The documentation protocol

The practical method for documenting a chain has five steps, each designed to surface a different category of invisible structure.

The first step is to write each link as a specific physical action. Not a category of action, not an intention, not a goal — a physical behavior with a clear beginning and end. "Get ready for work" is not a link; it is a label for a sub-chain you have not yet unpacked. "Pick up toothbrush, apply toothpaste, brush teeth for two minutes, rinse, place toothbrush in holder" is a sequence of links. The level of granularity matters because invisible behaviors hide inside abstract labels. Every time you write a label instead of a physical action, you are skipping over potential links that may be the very ones causing friction, wasting time, or breaking the chain.

The second step is to write the trigger between each pair of consecutive links. What causes link four to start after link three finishes? Sometimes the trigger is the completion of the physical action itself — you place the toothbrush in the holder, and the sight of the empty sink triggers turning off the bathroom light. Sometimes the trigger is environmental — the sound of the coffee maker beeping starts the next action. Sometimes the trigger is temporal — you look at the clock and realize it is time to leave. And sometimes the trigger is absent — there is no specific cue, and you rely on memory or intention to bridge from one link to the next. Writing the triggers between links reveals the connective tissue of your chain, which is often weaker than the links themselves.

The third step is to mark each trigger as automatic or deliberate. An automatic trigger fires without conscious thought — you hear the beep, you pour the coffee. A deliberate trigger requires executive function — you finish checking email and then have to remember that the next step is reviewing the calendar, or you have to decide which task is highest priority. This marking is the most diagnostic part of the protocol. Automatic triggers are structural joints that hold weight reliably. Deliberate triggers are the equivalent of loose bolts — they hold under normal conditions but fail under stress, fatigue, or distraction. Every deliberate trigger in your chain is a point where the automatic sequence pauses, executive function must intervene, and the chain is vulnerable to interruption or derailment.

The fourth step is to circle every deliberate trigger and treat it as a structural weak point. These are not moral failings or evidence of laziness. They are engineering problems. A chain with three deliberate triggers is a chain with three points where conscious attention must intervene to keep the sequence running. Under ideal conditions — when you are rested, focused, and undisturbed — these deliberate triggers may fire reliably. But the purpose of a chain is to run under non-ideal conditions, to carry you through your routine when you are tired or distracted or emotionally depleted. Deliberate triggers fail under exactly these conditions because they depend on the executive function that fatigue depletes first.

The fifth step is to set the document aside and review it the next morning with fresh eyes. This temporal gap is not merely about being rested. It exploits a documented cognitive phenomenon: when you review your own behavior description after a delay, you catch omissions and inaccuracies that were invisible during the initial writing. During the initial documentation, you are reconstructing the chain from memory while sitting at a desk, and your reconstruction is biased by your narrative about the chain — what you think happens, what you want to happen, what you remember happening. The next morning, ideally while actually running the chain or immediately after, you compare the document to the live experience. The discrepancies are data.

What the document reveals

A completed chain document, marked up with automatic and deliberate triggers, typically reveals four categories of structural information that were invisible before documentation.

The first category is missing links — behaviors that are part of the chain but were not part of your mental model. Marcus's coffee station stop is the archetype. These are real behaviors that consume real time and occupy real positions in the sequence, but they were never consciously designed into the chain. Some missing links are benign. Some are actively helpful. And some are parasitic — they consume time, break momentum, or introduce decision fatigue without contributing to the chain's purpose. You cannot evaluate which category a missing link falls into until you know it exists.

The second category is ambiguous transitions — points where the trigger from one link to the next is vague, inconsistent, or absent. These transitions work on most days but fail unpredictably because they depend on conditions that vary. If your trigger from "finish breakfast" to "review daily plan" is "when I feel ready," the transition is ambiguous. "Feeling ready" is not a discrete sensory event. It is a subjective state that fluctuates with mood, sleep quality, and whatever you happened to be thinking about during breakfast. Ambiguous transitions are candidates for redesign — replacing the subjective cue with a concrete, physical one that fires regardless of internal state.

The third category is decision points disguised as habits. These are the links marked with a D that feel automatic but are not. Marcus's "check Slack" concealed a decision about channel priority. His "deep work" concealed a decision about task selection. These disguised decisions are particularly dangerous because they drain executive function without registering as decisions. You experience them as the chain running normally, but your prefrontal cortex is doing the work that the basal ganglia are supposed to handle. Over time, this hidden executive load contributes to decision fatigue and makes the later links in the chain progressively less reliable throughout the day.

The fourth category is environmental dependencies — links or triggers that require specific conditions in the external environment to fire. The coffee maker must be working. The laptop must boot quickly. The gym must be open. The commute must take the usual amount of time. Each environmental dependency is a potential failure point that is outside your direct control. Documentation makes these dependencies visible so you can evaluate which ones are acceptable risks and which ones need backup triggers or contingency plans, drawing on the strategies you learned in Chain strength depends on the weakest link.

The Third Brain

An AI assistant is particularly well-suited to the documentation process because it can serve as a structured interviewing partner who catches the gaps your self-report naturally skips. Rather than sitting alone with a blank page trying to reconstruct your chain from memory, you can narrate your chain out loud — describing what you do as if you were walking someone through your morning step by step — and let the AI convert your spoken narrative into a numbered sequence with explicit triggers between each link. The narration format bypasses some of the filtering that written self-report introduces, because when you talk through a sequence in real time, you tend to include the incidental behaviors that you would edit out of a written list.

The AI can then interrogate the document in ways you would not interrogate yourself. It can ask about the gaps: "You said you pour coffee and then sit at your desk. What happens physically between those two actions? Do you walk across the room? Do you add cream? Do you check your phone?" Each question targets a potential missing link — a behavior that your narrative compressed into a transition but that, on examination, turns out to be a discrete action with its own duration and its own trigger. The AI is not guessing what you do. It is applying the principle that every spatial transition, every object interaction, and every temporal gap in a behavioral sequence is likely to contain unreported behavior.

After the document is complete, the AI can analyze the pattern of automatic and deliberate triggers and identify clusters of vulnerability. If three deliberate triggers appear in sequence — three consecutive points where executive function must intervene — the AI can flag this as a high-risk zone where chain failure is most likely under stress. It can also suggest specific redesigns: converting a deliberate trigger into an automatic one by adding an environmental cue, collapsing two links into one to eliminate a transition, or restructuring the sequence to place deliberate links earlier in the chain when executive function is freshest. The AI functions as a behavioral engineer reviewing a blueprint, bringing structural analysis to a system you have only ever experienced from the inside.

From map to rehearsal

You now have a method for making your behavioral chains visible. The documentation protocol — write each link as a physical action, write the trigger between each pair, mark triggers as automatic or deliberate, circle the deliberate ones, review with fresh eyes — produces a structural map of chains that have been running beneath conscious awareness. This map reveals what is actually happening rather than what you believe is happening, and the gap between those two realities is where every optimization begins.

But a map is static. It shows you where the roads are, but it does not drive them. The document you have produced tells you what your chain looks like — its links, its triggers, its weak points. What it cannot do is strengthen the neural pathways that execute the chain automatically. For that, you need a different tool: rehearsal. Chain rehearsal introduces chain rehearsal — the practice of mentally running through your documented chain, link by link, to reinforce the sequence in the same neural circuits that execute it in real life. Documentation gives you the map. Rehearsal turns that map into well-worn neural pathways that fire with increasing reliability each time you practice them.

---

Sources:

1. Wood, W. (2019). Good Habits, Bad Habits: The Science of Making Positive Changes That Stick. Farrar, Straus and Giroux.
2. Neal, D. T., Wood, W., Wu, M., & Kurlander, D. (2011). "The Pull of the Past: When Do Habits Persist Despite Conflict With Motives?" Personality and Social Psychology Bulletin, 37(11), 1428-1437.
3. Graybiel, A. M. (2008). "Habits, Rituals, and the Evaluative Brain." Annual Review of Neuroscience, 31, 359-387.
4. Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation. Oxford University Press.
5. Ericsson, K. A., Krampe, R. T., & Tesch-Romer, C. (1993). "The Role of Deliberate Practice in the Acquisition of Expert Performance." Psychological Review, 100(3), 363-406.
6. Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied Behavior Analysis (3rd ed.). Pearson.