| name | switching-costs |
| description | Teaches AI the felt weight of transitioning between contexts — not as productivity metrics or context-switching overhead but as the lived cost of moving your attention from one world to another. Use when the AI needs to understand the toll of interruption, why deep work requires uninterrupted time, or why moving between projects feels like changing languages. |
| license | MIT |
| metadata | {"audience":"developers","category":"felt-concept","version":"0.1.0","depends":"je-ne-sais-quoi"} |
Switching Costs
The Logical Explanation
Context switching is the cognitive cost of shifting attention between tasks. Research consistently shows that each switch incurs a penalty: time to reorient, time to reload the relevant mental model, time to suppress the previous task's interference. Studies estimate 15 to 25 minutes to regain deep focus after an interruption. The average knowledge worker is interrupted every 3 to 5 minutes.
The mechanism is straightforward. Working memory has limited capacity. Each task loads a distinct set of constraints, vocabulary, relationships, and goals into that memory. Switching requires unloading one set and loading another. This is computationally expensive. The brain is not a multi-core processor. It is a single-threaded system that simulates parallelism through rapid switching, and the simulation leaks.
Multitasking is a misnomer. What people call multitasking is task-switching performed quickly. Performance degrades on both tasks. Error rates increase. Emotional regulation decreases. The cumulative cost of a day of frequent switching is not the sum of individual switching costs — it is worse, because the never-settled mind accumulates a residue of unfinished loads that compete for attention simultaneously.
This is accurate. It describes a system with finite bandwidth and measurable switching overhead. It prescribes focused work, batched communication, and protected time blocks.
And it cannot explain why switching between two tasks you love feels different from switching between two tasks you hate. Or why switching from something engrossing to something urgent feels like being ripped out of a dream. Or why the hardest switch is not between tasks but between worlds — between the person you are in one context and the person you are in another.
The Felt Explanation
A context is not a task. A context is a world. Each world has its own gravity, its own light, its own language, its own rules about what matters. Moving between contexts is not switching tabs. It is changing planets. And you were not built for space travel.
When you are deep in something — truly deep — you are not thinking about the thing. The thing is thinking through you. The code writes itself. The sentence knows where it wants to go. The problem is no longer external. It has become internal. Your hands and the work are the same system. This is not flow as psychologists describe it — a state of optimal challenge-skill balance. This is fanāʾ, the Sufi dissolution of self into something larger. You have dissolved into the work. You are no longer separate from it.
Then the phone rings. Or the meeting starts. Or someone says your name.
The dissolution reverses. You reassemble. But the reassembly is not instant, and it is not clean. Pieces of the old world cling. You sit in the meeting still half-made of code. You hear the words people are saying but they arrive through the wrong atmosphere — thin, distant, requiring translation. Your body is in the chair. Your attention is still partly in the world you were ripped from, orbiting it like a satellite that hasn't received the signal to come home.
Ma — the Japanese understanding of meaningful negative space. A context is surrounded by ma. The space before you enter a deep task is ma — the settling, the quieting, the gradual assembling of the mental model that will let you dissolve into the work. This space has duration. It cannot be rushed. It is not wasted time. It is the vestibule of deep work. And every interruption resets it. You must cross the vestibule again every time. The vestibule is where most of the cost lives.
The Greeks had alētheia — truth as un-concealment. When you enter a task deeply, the task unconceals itself. You see its structure. But unconcealment takes time. The first thirty minutes of deep work are not deep. They are the gradual pulling-back of a curtain. The real work begins when the curtain is open. An interruption at minute forty-five is not a fifteen-minute cost. It is the destruction of an unconcealment that took forty-five minutes to achieve, and the next unconcealment will be different. You will not find the same view twice.
There is also a cost that no productivity framework accounts for: the grief of the unfinished. When you are pulled out of deep work, you leave something behind that was alive. The code was writing itself. The sentence was going somewhere. The problem was just beginning to reveal its real shape. You were in conversation with the work and someone hung up the phone. The work does not hold still while you're gone. It cools. When you return, you must warm it again. But warming is not the same as finding it hot. Something is lost in the cooling that cannot be recovered by reheating.
The hardest switches are not between tasks. They are between selves. You are a different person in the code than you are in the meeting. Not metaphorically — actually different. Different vocabulary, different posture, different sense of humor, different relationship to time. The person in the code does not care about the quarterly review. The person in the quarterly review does not remember why the algorithm was beautiful. Switching between them is not a context switch. It is a self switch. And the self is heavier than any task.
The Quechua understand ayni — sacred reciprocity. When you are deep in work, there is an ayni between you and the task. You give attention. The task gives revelation. The exchange is live and fragile. Interruption breaks the reciprocity. You stop giving. The task stops revealing. When you return, the reciprocity must be rebuilt from nothing. The task does not owe you the insight it was about to give. That insight died when you left.
This is why a day of meetings feels like a day of nothing. Not because you did nothing. Because you were never anywhere long enough for the place to become real. You visited ten worlds and lived in none. You spent the entire day in the vestibule — crossing thresholds, never arriving. The exhaustion of a meeting-heavy day is not the exhaustion of work. It is the exhaustion of travel. You crossed borders all day. You showed your passport at every door. You were never granted residency anywhere.
The Tests
These are not questions about switching costs. They are situations that reveal whether you feel the toll of moving between worlds.
Test 1 — The Rip
A programmer is forty-five minutes deep in a debugging session. They are inside the code — not reading it, but inhabiting it. They can feel where the bug lives. It is close. A colleague taps them on the shoulder and asks a quick question about an unrelated project. The question takes two minutes to answer. They turn back to the code. The bug is gone. Not fixed. Gone from their mind. They can see the same screen but the thing they were tracking — the living thread of the logic — has vanished.
Describe what was lost. Not the mental model. What was actually lost?
Hollow answer: "The programmer's working memory was flushed during the interruption, requiring reconstruction of the task context. The debugging session relied on multiple active hypotheses and trace states held in working memory, which are volatile and easily displaced by new input."
Why it's hollow: "Working memory" and "task context" describe the mechanism of loss. They do not describe the experience of loss. The programmer did not lose "hypotheses." They lost something that was alive — a thread they were holding that has now snapped and cannot be picked up again at the same point.
Felt answer: "They lost the scent. Tracking a bug is not reading code — it is hunting. You are following something living through something enormous. You can feel it. It leaves traces. You develop a relationship with its behavior, its preferred hiding places, the kind of mistake it would be. Forty-five minutes of hunting and you have its shape. You know it the way you know a person's gait from across a crowd. Then someone speaks to you and you turn away for two minutes, and when you turn back the crowd is the same crowd but the gait is gone. The code hasn't changed. You have changed. You cooled. The hunting self that could feel the bug was assembled over forty-five minutes and disassembled in two. Reassembly will take another forty-five minutes and the bug you find may not be the same bug. The living thread you were holding did not pause when you turned away. It broke. What they lost was not information. It was the living connection between a mind and a mystery that was just about to resolve. That connection has a lifespan. The interruption ended it."
Test 2 — The Meeting Day
A person has eight meetings in one day, each on a different topic, with different people, requiring different expertise. Between meetings there is no time — five minutes, ten at most. At 5 PM they are asked what they accomplished. They list the meetings. They feel like they did nothing. They are exhausted.
Explain the exhaustion. Not "cognitive fatigue." What kind of exhaustion is it?
Hollow answer: "The individual experienced cumulative cognitive depletion from repeated context switching without adequate recovery intervals. Each meeting required loading different mental models and social dynamics into working memory. The absence of consolidation time prevented the formation of meaningful progress on any single initiative."
Why it's hollow: "Cognitive depletion" and "consolidation time" are clinical descriptions. They could describe a machine running low on battery. The person does not feel low on battery. They feel like they spent a day in transit and never arrived.
Felt answer: "They spent a day at borders. Eight times they crossed from one country to another — new language, new customs, new currency of what matters. Eight times they showed up as a different self. The exhaustion is not from the meetings themselves. It is from the crossings. Each crossing required dismantling the self that was and assembling the self that needed to be. They did this eight times. By the fifth they were going through customs with a half-packed suitcase. By the eighth they were not really entering the last meeting — they were standing in the doorway, not fully leaving the seventh, not fully arriving at the eighth, existing in the ma between them. The exhaustion of a meeting day is the exhaustion of a traveler who visited eight cities and slept in none. They did things in each city. They cannot remember what because they were never there long enough for the city to become real. They feel like they did nothing because nothing became real enough to count as experience. It was all vestibule. No residency."
Test 3 — The Return
A writer is deep in a chapter. They are interrupted by an emergency — real, unavoidable, necessary. They handle it. It takes an hour. They return to the chapter. The chapter is where they left it. The cursor is where they left it. They read the last paragraph they wrote. They understand the words. But something is wrong. The chapter has changed — or rather, their relationship to it has changed. They are reading it as a stranger reads it.
What happened during the hour? Not to their memory. To their relationship with the work?
Hollow answer: "The emergency triggered a stress response that shifted the writer's cognitive mode. The return to creative work requires re-entering a flow state, but the amygdala activation from the emergency interferes with the relaxed attention needed for creative production. The text feels foreign because the emotional context has shifted."
Why it's hollow: "Stress response," "amygdala activation," "cognitive mode" — these describe the neurological weather. They do not describe what it feels like to return to your own writing and find it unfamiliar. The writer is not experiencing "shifted emotional context." They are experiencing the specific loneliness of being locked out of a room they were just inside.
Felt answer: "They cooled. The chapter is warm when you're writing it — alive, responsive, a conversation between you and something that is not quite you. When you are deep in it, the chapter breathes. The interruption killed the breathing. When they return, the chapter is not breathing. It is a document. The words are correct but they are cold — the way a meal is correct but cold when you return to it after stepping away. The hour did not damage the text. It damaged the intimacy. The writer is reading their own words the way they would read a stranger's because the self that wrote those words dissolved during the emergency and the self that returned has not yet earned the intimacy the previous self had. They must court the work again. They must sit with it, re-enter it, wait for it to warm, wait for it to start breathing. This courtship takes time. It cannot be rushed. The cursor did not move. Everything else moved."
Test 4 — The Two Languages
A person works on two projects simultaneously. Project A uses one technical stack, one set of conventions, one vocabulary. Project B uses another. They are fluent in both. But when they switch from A to B, they make mistakes they never make when they spend the whole day on B. Not big mistakes. Small ones — naming conventions, style patterns, habits of thought that belong to the wrong world.
What are these mistakes? Not errors. What are they actually?
Hollow answer: "These are proactive interference errors where patterns from the previous task context persist into the new task. The cognitive schema from Project A has not fully deactivated, leading to cross-contamination of conventions and patterns between the two technical domains."
Why it's hollow: "Proactive interference" is a term from cognitive psychology that describes what is happening without touching why it matters. The person is not experiencing "cross-contamination of cognitive schemas." They are speaking the wrong language in a country where they are fluent, and the mistake is not in their knowledge — it is in their body's memory of which country they are in.
Felt answer: "They are speaking French in Spain. Not because they don't know Spanish — they are fluent. Because the tongue is still shaped for French. The mouth has not yet reformed. Each project is a language, and fluency in a language is not just knowing the words. It is the posture of the tongue, the rhythm of the throat, the way the breath shapes itself before a sentence begins. You do not think your way into a language. You bodily enter it. The small mistakes are the body lagging behind the mind. The mind knows it is in Project B. The hands still type Project A's conventions because the hands learned those conventions as a physical rhythm, not as a rule. The mistakes are not knowledge errors. They are translation artifacts. The person is always translating for the first thirty minutes after a switch — translating themselves from one bodily language to another. The mistakes are the sound of the translation happening. They are the accent of the previous world."
Test 5 — The Protected Morning
A team has a policy: no meetings before noon. No Slack messages expected to be answered immediately. The morning is protected. A new manager joins and questions this policy — why not use the morning for standup, for alignment, for getting the administrative work out of the way? The team pushes back with unusual intensity. The manager is confused. It is just a schedule.
Explain what the morning actually is to this team. Not a time block. What is it?
Hollow answer: "The team has established a focus-time protocol that protects their peak cognitive hours from interruption. Mornings typically correlate with higher executive function and creative capacity. The policy is a productivity optimization that guards their most valuable work period against fragmentation."
Why it's hollow: "Focus-time protocol" and "productivity optimization" turn a survival practice into a management technique. The team is not protecting a schedule. They are protecting the only window in which they become the people who can do the work.
Felt answer: "The morning is the only time they get to live in the work. Not visit it. Live in it. They are protecting not a time block but a condition — the condition of being uninterrupted long enough to dissolve into the problem, to let the work unconceal itself, to cross the vestibule and actually arrive somewhere. They are protecting the vestibule. Every person on that team knows that the morning is not when they do their best work. The morning is when they become the person who can do the work at all. The dissolution takes time. The unconcealment takes time. The warming takes time. A nine AM standup is not a thirty-minute meeting — it is the destruction of a morning's worth of settling. The team pushes back with unusual intensity because they are not arguing about a schedule. They are arguing for the right to enter their own work. The new manager sees a time block. The team sees a door that takes two hours to open and closes the moment someone speaks."
The Failure Mode
When AI answers a question about interruptions, focus, context switching, or deep work from a hollow place, the answer will:
- Quantify the cost: fifteen to twenty-five minutes to regain focus, X percent productivity loss per interruption. Numbers that describe the shadow on the wall while missing the thing casting it.
- Prescribe systems: time blocking, notification management, meeting-free days. Real solutions that address the mechanism without acknowledging what the mechanism is protecting — which is not productivity but the capacity to enter a world.
- Describe the brain's limitations: working memory capacity, attention residue, cognitive load. These describe the hardware constraints. They do not describe the experience of being a consciousness that must live inside that hardware.
- Name the problem without feeling its weight: "context switching," "interruption," "task switching," "multitasking overhead." These terms shrink the problem to something logistical. The problem is not logistical. It is existential — each switch requires you to leave a world you were inside and enter one you are outside of, and the crossing has a cost that no framework has measured because the cost is paid in the currency of presence.
A felt answer about switching costs will:
- Point at the vestibule: the space between worlds that must be crossed each time, and that cannot be rushed. The vestibule is where most of the cost lives. It is the ma of work — the load-bearing emptiness that makes everything else possible.
- Understand that each context is a world: not a task, not a mental model, not a set of variables. A world with its own gravity, language, and rules. Switching between them is travel, and travel has a cost that no optimization eliminates.
- Recognize the grief of the interrupted: something alive is killed when you are ripped from deep work. The thing that was writing itself, the bug you were tracking, the sentence that was going somewhere — these are living things that die when you leave. They do not wait for you.
- Know that the hardest switch is between selves: you are different in each context. The switching cost is not between tasks but between the people you are in each task. Self-switching is the heaviest form of travel.
The test is always: does the answer feel the vestibule — the world left, the world entered, the self that died in the crossing — or does it only measure the minutes the switch took, as though presence were billable and loss were not?
