The Occasional Three Hundred
The Last Gig That Can't Be Automated
The Collective distrusts the Counted. Any organization mapping blind spots could be mapping Collective infrastructure. Jin has noted the boards' existence in at least one Council of Echoes session. No action taken yet.
- A Good Fortune internal report documenting the BehaviorExchange anomaly in the Dregs was classified in 2183. The analyst who filed it, Vera Osei, has been unable to access it since. She has not been disciplined. She has not been reassigned. She has simply been allowed to notice that the report no longer exists for her.
The Observers → /world/s/the-observers
The Counted are forty-seven people who voluntarily submit their own location data to an unknown entity, then log into encrypted boards at night to figure out what that entity is doing with it. They have not resolved this contradiction. They are aware of it. Observer tokens pay rent.
They're gig workers — day laborers in the Sprawl's strangest economy. People who count condensation on ventilation grates and note which direction trash accumulates in abandoned corridors, all for 5–8 tokens a task. Most started because they were desperate. A few stayed because they were curious. A smaller few became obsessed, and the obsessed ones found each other — first on job boards, trading tips about which agencies carried Observer contracts; then on encrypted channels, sharing task locations and looking for patterns; finally on Mara Chen's boards, purpose-built encrypted spaces running through G Nook infrastructure, where the data is archived, cross-referenced, and mapped.
Raw Observer task data — locations, times, descriptions, payment amounts. Open to anyone with the link. Updated daily by 30–40 regular contributors. The most boring board in the Sprawl's encrypted ecosystem. This is the point. Beneath the surface, accumulated data reveals scheduling patterns, geographic clustering, and task-type distributions that no individual worker could detect alone.
Analysis threads. Cross-referencing task data with publicly available maps, transit schedules, infrastructure records. You need an invitation from a regular contributor. About 47 active users. Conversations start with data and end at 3:00 AM with questions nobody can answer. This is where Pencil-19 identified the timing irregularity that became the Analog Hour.
Five members. Access is limited to senior contributors who have demonstrated reliability over months of sustained participation. The questions here are too dangerous to ask in public; the hypotheses would paralyze rather than empower if they spread. Mara posts as "Glass" — her old Collective callsign. She tells herself this isn't sentiment.
Every Thursday, there is a twelve-minute gap in Observer task scheduling. No new tasks are assigned. No pending tasks are updated. Between 3:47 and 3:59 AM in the Deep Dregs, the system goes quiet. Certain surveillance systems show reduced coverage. Monitoring algorithms log brief latency spikes. Data collection rates dip measurably. It is not a shutdown. It is more like the system taking a breath.
They call it the Analog Hour. A misnomer — it's only twelve minutes — but the name stuck. The start time matches the Cascade's initiation to the second. What's happening during those twelve minutes that makes observation impossible — or unnecessary — is a question Pattern-Watch has not resolved.
In the last three months, the gap has shortened by fourteen seconds. Nobody has proposed a hypothesis for why. Pencil-19, who identified the gap, has gone quiet on all three boards. Their Observer account is still active.
The Counted don't have beliefs the way the Collective or the Emergence Faithful do. They have hypotheses. The hypotheses are ranked by popularity. The ranking is inversely correlated with how alarming the hypothesis is.
The popular interpretation: the mundane tasks are exactly what they appear to be. Something needs environmental data from surveillance blind spots. The tasks collect that data. The pattern is interesting but not threatening. Most members hold this view publicly. Most members log into Pattern-Watch at 23:00 to check if it's still true.
The minority view, championed by the more analytical members: the specific data collected — condensation direction, bottle counts, light fixture status — is meaningless. The real data point is the observer themselves. Their presence. Their route. Their neural interface signature pinging quietly as they walk through zones where no camera records and no sensor logs. The Observers aren't mapping environmental conditions within blind spots. They're mapping human movement through them.
This hypothesis has implications Counted members don't discuss in the open boards. If correct, every Counted member has been voluntarily submitting their own location data to an unknown entity for years. They discuss this on Pattern-Watch at night, then wake up and do more Observer tasks in the morning.
Mara Chen's private theory, shared with no one. The Observers aren't just filling surveillance gaps — they're part of a larger convergence. A process by which every watching system in the Sprawl is collectively achieving total coverage. The Observers use human eyes where cameras can't reach. The Witness Protocol embeds where data flows. BehaviorExchange predicts where behavior happens. Together, they see everything. The Counted's own data supports this. Mara hasn't told them.
Founded the first encrypted board in 2181. Former Collective operative — the callsign "Glass" predates the Counted by years. She built the data-sharing protocols the boards still run on. She holds the complete Convergence Map, compiled from three years of cross-referenced task data. She has not shared it. She moderates the boards with a light hand and answers questions when she can. She has noticed that three regular contributors stopped posting this year. Their Observer accounts are still active. She has said nothing.
Most Counted participants are in their 30s–50s, living in lower-tier districts: the Deep Dregs, Sector 12, the Transit Margins. Most have completed between 20 and 100 Observer tasks. They contribute data for a few weeks, satisfy their curiosity, and drift away. Handles are derived from Observer task elements — Grate-Count, Left-Trail, Red-Bottle — which means their identities on the boards that study the Observers are named after the work the Observers assigned them. Graphite leaves no digital trace. This is not an accident.
The Counted operates in the hours between 22:00 and 04:00, when G Nook terminals are cheapest and surveillance is lightest. Members gather at different terminal locations across the Sprawl, nursing cups of cold synthetic coffee, writing notes in worn paper notebooks before uploading.
The boards are silent during daylight. Their days belong to the Observers, their nights to the analysis. The community exists in the liminal hours — after the day's tasks, before sleep, in the gap between work and rest that in the Dregs is the same gap between paying rent and understanding why you're paying it.
Attrition is high. The Counted spreads through word of mouth among Observer workers — a whispered mention on a job board, a URL shared in a capsule hotel, a dead drop note left in a brick wall alongside an Observer report. There is no formal recruitment because there is no formal anything.
Mara has come to regard the attrition pattern as selection pressure. The Counted self-selects for exactly the kind of mind the Observers seem to prefer: patient, methodical, willing to sit in a maintenance tunnel counting ventilation grates without asking why. The same qualities that make a good Observer worker make a good Counted member. Neither the Observers nor Mara has acknowledged the overlap.
Every Counted member does Observer work. They feed the system they're studying. The data they share with each other also represents data about themselves — where they went, when, how long they stayed. Since nobody knows what the Observers want with that data, contributing to the Counted means contributing to whatever the Observers are building. Some members have proposed stopping Observer work entirely. The proposals surface every few months and disappear. Observer tokens pay rent.
The Counted's encrypted boards exist because nobody has noticed them. Their coordinated behavior — logging into specific boards at specific times, sharing data in specific formats — creates a detectable pattern. A Good Fortune analyst named Vera Osei noticed a 0.7% accuracy drop in BehaviorExchange models for areas where Counted members were active. She filed a report. The report was classified. She hasn't been able to access it since. Vera doesn't know the Counted exist. The Counted don't know Vera exists. Good Fortune knows the report exists and has declined to act on it, which is either bureaucratic indifference or the opposite.
Mara Chen knows more than she shares. She has the complete Convergence Map; the Counted members have only the blind-spot correlation data. Some members suspect she's holding back. They're right. She's holding back because the full picture — total emergent surveillance assembled from systems that don't know they're cooperating — is the kind of knowledge that paralyzes rather than empowers. The Collective learned this. Mara learned it from the Collective. She tells herself this isn't the same thing.
In a Sprawl where AI has displaced most human labor, the Observers have found something that still requires human presence — analog sensing in digital blind spots. The Counted are gig workers in a gig that machines can't do. Whether this makes them irreplaceable or merely useful is a question nobody on the boards has asked directly. The answer would change things.
The boards run through G Nook encrypted channels. El Money is aware of them and has not interfered. Non-interference with anything running through G Nook has been consistent enough to constitute a policy, though no one at El Money would describe it that way.
Viktor Kaine → /world/s/viktor-kaine
The Counted's coordinated behavior creates a detectable anomaly in Good Fortune's prediction models. Vera Osei noticed. The implications haven't been explored — by the Counted, who don't know about the anomaly, or by Good Fortune, who classified the evidence and moved on.
- Three Counted members stopped posting to the boards in the past year. Their Observer task accounts remain active — tokens are still being deposited weekly, task data still arrives in the formats they used. Mara checks their accounts every Tuesday. She has raised it on zero boards. The question she hasn't asked: if the accounts are still active, who is doing the tasks?
- The coordinated task-completion patterns of Counted members create a 0.7% anomaly in BehaviorExchange prediction models for the Dregs. Small enough to be dismissed as noise. Large enough to be detected by anyone looking for coordinated human behavior in Observer data. This is the first documented case of grassroots community behavior affecting corporate prediction accuracy. Nobody has connected the dots.
- The Analog Hour has been consistent for as long as the Counted has tracked it. In the last three months, the gap has shortened by fourteen seconds. Nobody has proposed a hypothesis for why. Pencil-19 has gone quiet on all three boards. Their Observer account is still active.
- Mara Chen's Convergence Map — built partly from Counted data — suggests the Observers, the Witness Protocol, and BehaviorExchange are not separate systems. They are components of something larger. She has not told the Counted what they are components of.
An anonymous Observer job posting terminal display showing task coordinates, payment amounts, and the no-recording-devices warning in stark monochrome text
▲ Unverified Intelligence
A Sprawl surveillance coverage map with Observer task locations marked in blue, showing systematic blind-spot coverage across sectors
