Northstar passes one million, unemployment rises

On 27 September Northstar, the city in the CS2 project, passed one million actually simulated residents. Supply and the job market did not keep pace throughout: between the first and last available readings, unemployment rose from about 11 to about 31 percent. The day also shows how human priorities, execution by the building agent and later checks fit together, including the correction of wrong decisions.

Growth is documented, lasting stability is not

The reporting period is 27 September in Berlin time; all measurement times below are UTC. The day's data file holds 50 population readings, the first shortly after the Berlin day began, the last shortly before it ended.

Between the two population endpoints lie about 460,000 additional residents. That is a difference between observed stocks, not an exact midnight balance and not a count of all people moving in and out. Unemployment is recorded at only 34 of the 50 points.

A separately stored game-state response already shows more than one million residents at 21:47 UTC, so the threshold was demonstrably crossed before the day ended. That reading proves neither the exact moment of the first crossing nor that the population stays above the threshold afterwards.

Human requirements, agent execution

The day was not purely autonomous expansion. The work records explicitly document human requirements: faster growth, more varied buildings and power lines routed underground, as well as infrastructure reserves and dealing with visible supply problems. The building agent carried them out.

In the northeast district, the design request led to a mix of residential and commercial buildings, medium density, terraced houses and detached houses. Readbacks of 23 rezoning sections add up to nearly 3,300 zone cells, with no building loss in the before-and-after comparisons. The figure describes designated area, not finished homes or additional residents, and the effect of this rezoning cannot be separated from the city's subsequent growth.

The work notes also record a temporary cut in residential tax and a renewed increase after the million was reached. Taxes, new zones and infrastructure changed at the same time; there is no controlled comparison of which intervention contributed how much to growth.

Underground power and its risks

One aim of moving the power routes was to remove disruptive terrain bulges. The building agent documented an inventory, the construction of deep replacement routes and a step-by-step switchover of connections. Its closing note describes an underground network, but also tolerances at old nodes and a technically limited exception at the external connection, so the requested depth is not met without exception.

The conversion did not run without disruption. According to the build log, an aborted rebuild temporarily caused power outages. In a later step an external connection was accidentally removed after an earlier error in a command sequence had not stopped the sequence. The building agent then loaded a saved state and repeated the affected steps with adjusted error handling, so progress on the day was not continuous.

Separately, the game crashed in the morning. The recorded diagnosis names a native runtime error; a causal link to the control interface has not been shown. Restoring from an autosave followed human approval. The day's series has a large gap between 06:27 and 09:55 UTC; crash and recovery explain the documented interruption, but not automatically every missing reading.

Supply: location and reach matter

While water supply was being expanded, new groundwater pumps were placed near a waste incinerator; according to the work records, the site check had not sufficiently accounted for polluting service buildings. The pumps were removed, replacement sites checked for ground pollution and the site evaluation extended. A stored full recheck afterwards contains no "Dirty Water" message and no related pump warning. That supports the fix of this incident, not the lasting safety of every water source; one conspicuous edge location remained under observation.

Late in the evening, the focus shifted to deathcare, rescue services and air pollution. The notes describe fully occupied vehicle fleets and long, congested access routes, particularly in the lake district, plus local air problems near an industrial strip. Additional facilities, new access roads and rezonings are documented as measures. The summary check of this work is dated 22:02 UTC, already 28 September in Berlin, so its reported drop in warnings does not count as a result of this day. Expected office jobs count as created only once readbacks show them.

More residents need more than building land

Rising unemployment is the clearest counterweight to population growth, and the unchanged satisfaction index at start and end masks this pressure. A high city-wide supply total likewise does not say whether a given district is reached in time; jobs, local service vehicles, routes and polluted sites have to be examined together.

The optimisation notes set out traffic and labour-market hypotheses and document the start of a metro service. A full line query at 21:28 UTC shows two metro lines. No drop in car traffic or CPU load follows from that, and with population, network, services and traffic all changing during the day, no isolated effect of any single measure has been shown.

For the regular performance series, qualified watcher readings are missing on 27 September, and the whole day is marked as a gap with unknown cause; other game and diagnostic records do not replace it. The report adopts neither the far-reaching projections in internal optimisation notes nor a guarantee that the long-term population target can be reached at constant simulation speed.

Status

At the end of the day Northstar has more than one million residents, but whether it stays there, and how the late supply measures work, is open. Results reported after the day belong in the next report.