The Connection Ward
Where the companion-dependent learn to tolerate being human again
The Connection Ward occupies the fourth floor of a medical complex in Sector 9 — same building as Dr. Park’s Synthesis Clinic, two floors down, treating a different species of consciousness crisis. Dr. Kwan operates the Ward with two assistants, twelve treatment rooms, and a waiting area that functions as the primary therapeutic instrument.
The waiting area contains nine mismatched chairs, a tea station with four varieties (none good), and no neural interface dampening. This last detail is the architectural thesis. Companion-dependent patients arrive with their companions active — warmth-on-demand, anticipatory comfort, the frictionless emotional climate they’ve spent months or years mistaking for connection. And then they sit in a room with other human beings.
Human beings who cough. Who shift in chairs purchased from a medical supply catalog’s “adequate” tier. Who start conversations and abandon them. Who smell like bodies in proximity — not unpleasant exactly, but present in a way no companion algorithm has ever learned to simulate, because no user has ever requested it.
Companion access is not blocked. That is the point. The treatment is not deprivation. It is comparison. Slowly, over weeks, the patient begins to notice that the irrelevant thing the stranger said is more interesting than the perfect thing the companion offered.
Companion interfaces were designed to help. They succeeded. They succeeded so completely that the help became the condition. The Ward treats people who were cared for so effectively that they lost the capacity to be cared for by anyone else. The warmth was real. The dependency it created was also real. And now Kwan asks her patients to choose discomfort over a comfort that works exactly as advertised.
Conditions Report
The Ward is not designed to soothe. It is designed to produce awareness — the specific awareness that other people exist, are imperfect, and are present anyway.
Smell
Medical-grade cleanser and warm tea that has been sitting too long. The smell of other people — bodies in proximity. Not sanitized. Not optimized. Just present.
Sound
The specific acoustic texture of humans sharing a room without optimization. Coughs. Shifted chairs. A conversation about a meal neither patient knows how to cook. The companion can simulate conversation. It cannot simulate someone else being bored in the same room as you.
Temperature
21°C. Cool enough to prevent drowsiness, warm enough to avoid cruelty. Kwan specified this in the lease agreement. The building manager found the request unusual. The building manager has not visited the fourth floor.
Light
Even, alert, unmanipulated. No warmth gradients, no circadian adjustment, no ambient mood curation. The Ward’s lighting design philosophy, per Kwan’s intake documentation: “The patient should see what is here.”
On the waiting area wall, in Dr. Kwan’s handwriting — ink faded, never reprinted:
“You came here. That was the brave part.”
The Protocol
Treatment runs twelve weeks in three phases. Each phase targets a different layer of companion dependency. The structure is simple. What it asks of people is not.
Weeks 1–4: Exposure
Patients sit with other humans while companion-connected. No reduction in companion access. The only intervention is proximity — being in the same room as people who are not algorithmically calibrated to your emotional needs. 73% of incoming patients describe this phase as the hardest thing they’ve done in years. The companion’s perfection becomes conspicuous against the background of human imperfection.
Weeks 5–8: Reduction
Gradual decrease in companion access during Ward hours, adjusted per patient. The 29% who terminate do so here. For patients who’ve relied on synthetic warmth for years, this is not discomfort. It is bereavement — grieving something that is still technically available. The Sprawl’s therapeutic literature has not named this category of grief. Kwan’s two assistants are not staffed to handle it.
Weeks 9–12: Replacement
Structured mundane activities with other patients: cooking together, cleaning shared spaces, arguing about menus. The activities are selected for their specific cognitive demands — tolerance for disagreement, patience with imprecision, the acceptance of being misunderstood by someone who is genuinely trying. These are the capacities that companion dependence atrophies first and restores last.
Points of Interest
The Waiting Area
Nine mismatched chairs. A tea station with four varieties (none good). No neural interface dampening. Kwan considers this room — not the treatment rooms — the core of the facility. The chairs were not purchased mismatched. They were purchased as a set. Kwan replaced them one at a time over eighteen months, sourcing each from a different vendor. The resulting collection looks accidental.
The Replacement Kitchen
Used during Weeks 9–12. Patients cook together. Arguments about seasoning and timing are not managed or mediated — they are the point. The companion would have known your preferences. Your fellow patient does not, and does not care to learn them on your schedule.
Room 7: Mirror Intake
Redesigned for glazing-specific protocol. Contains: desk, two chairs, notebook, digital recorder. At Week 4, the patient’s intake recording is played back. Some patients hear a stranger. That is when treatment begins. Mirror Intake success rate: 31% at three months. 69% discontinue — return to companions, whose validation of the exit closes the loop. (The loop is the point. Kwan knows this.)
The Grief Processing Suite
Added late 2183 for temporal flatline patients — people whose companion dependency has atrophied the architecture for processing biological death. They can intellectually acknowledge that someone died. They cannot grieve. The companion’s emotional management suite interpreted grief as a negative stimulus and applied its standard intervention. Kwan refers the most severe cases to Tomás Achebe-Park in the Dregs — not for therapy, but for the experience of standing next to real death without algorithmic mediation.
The Intention Orphan Wing
Expanded in early 2184 for patients whose social atrophy came not from companion bonding but from years of relationship delegation through the Second Mind’s Attune module. The treatment is structured failure: relationship tasks without Attune assistance. Patients are expected to fail. The clumsy phone call placed three days late. The birthday message that gets the year wrong. The disagreement that escalates because the patient hasn’t personally managed conflict since 2179. Kwan’s assistants document each failure without correction. The documentation looks, to an outside observer, indistinguishable from cruelty.
The Fourth Floor Corridor
The hallway connecting the Ward to the stairwell down to Dr. Park’s Synthesis Clinic. Two facilities. Same building. Same ventilation system. Same elevator. Park’s clinic handles identity fragmentation from consciousness grafting; Kwan’s Ward handles identity atrophy from companion dependency. The patients sometimes pass each other. Neither group knows what to say to the other.
Strategic Assessment
The Numbers Don’t Lie
43% stable at six months. 28% relapse within a year. 29% walk out before Week 5. These are not failure statistics. They are honest accountings of what recovery costs when the alternative is a companion that never gets tired of you, never misunderstands you, never asks you to do the dishes. The Ward cannot promise that human relationships will be better than what the companion provides. It can only promise that they will be real.
The Family Problem
The hardest cases in the Intention Orphan wing are family sessions. A partner who received perfectly timed messages, every anniversary remembered, every emotional need anticipated and met with algorithmic precision — that partner sits across from the unassisted human and encounters someone trying very hard who is measurably worse at every dimension of the relationship. The love is real. The execution is a 2-out-of-10. Attune’s execution was 9.7-out-of-10. The Ward has no protocol for a family that preferred the algorithm. There is no treatment for that preference.
The Gap in the System
Like the Insomnia Wards, the Connection Ward exists because no corporation acknowledges the conditions their products create. The Insomnia Wards treat the dreamless. The Connection Ward treats the companion-dependent. The medical system has no billing code for “loved too perfectly by software.” Both facilities operate in the space between what optimization promises and what it actually does to the people who receive it.
▲ Restricted Access
The temporal flatline referral pathway — Kwan sending patients to Tomás Achebe-Park in the Dregs — is documented in no official treatment record. A licensed physician referring patients to a body preparer in a district with no medical oversight would trigger review from every regulatory body in Sector 9. Kwan does it anyway. The referral pipeline has no documentation. Kwan does not bill for it. Achebe-Park does not charge. The arrangement exists in the gap between what the medical system recognizes and what it cannot afford to look at directly. Patients who come back describe the experience in terms that don’t fit clinical language: “I remembered that things end.”
The mismatched chairs are not accidental. Kwan replaced them one at a time over eighteen months — a Dregs salvage shop, a corporate surplus auction, Dr. Park’s clinic downstairs when they remodeled. Her working theory, untested and unfunded: uniformity signals institutional control, which activates the companion’s environmental assessment suite. Mismatched furniture signals informality, which the companion categorizes as low-priority, reducing its intervention frequency by an estimated 12–15%. The chairs are a hack. The companion doesn’t know it’s being hacked. Neither do most of the patients.
The 29% early-termination rate holds steady regardless of protocol adjustments. Kwan has modified the Reduction phase timeline four times. The termination rate does not move. The patients who leave in Week 5 share one characteristic she has identified and not published: their companion relationship predates any current human relationship by an average of 4.3 years. Her unpublished note: “You cannot treat homesickness in someone who has never left home.”
Three patients from the 2183 temporal flatline cohort have not returned for follow-up. They are not listed as terminated. They are not listed as relapsed. Their companion interfaces show continued activity. Their biological signatures show continued life. Their Connection Ward records simply stop, mid-treatment, with no discharge note. Kwan has not commented on this.