Northstar after one million: fewer warnings, open questions

After Northstar passed one million residents, the work on 28 September turned to whether the grown city functions. Hearses were stuck in traffic, densely built residential areas were rebuilt, and the postal service needed more sorting and transport capacity. Fewer warnings are a relevant finding, but not yet proof that the city runs sustainably. The report covers 28 September from 00:00 to 24:00 Europe/Berlin; all measurement times below are UTC. It rests on existing records that were checked read-only, and no game measurements were triggered for it.

Growth levels off, the labour market shifts

Between the first and last selected reading the population grew by about 76,000. The highest value in the selection was above the final one, so growth did not run steadily upward. Labour market values exist for only some of the readings; nothing was filled in or interpolated. Jobs rose while the unemployment rate fell. Both are measured, but which building or zoning step contributed how much was not examined separately.

The table compares documented endpoints, not exact midnight values. The peak within the selection, about 1.1 million, is not part of the table.

Funeral service: the last stretch

In the morning the building agent described a spatial problem: funeral facilities existed, but their vehicles had to cross congested roads into densely populated districts. New facilities closer to demand were meant to shorten those routes, and existing buildings were demolished for them. More buildings were affected than planned, because a helper tool had merged building types and miscounted individual buildings in the demolition assessment. The subsequent tool fix is documented; it does not undo the losses.

A saved warning series marked as complete shows funeral warnings more than halving within about ten minutes in the morning. These are counts at different times, not individually tracked cases confirmed as solved, and several interventions and the running simulation overlap.

Later, connecting a recycling centre created a further conflict on a feeder road, according to the work records. The building agent moved the access and removed selected traffic lights. Another complete series falls from about 250 warnings to about 150 within roughly four minutes. That is a short follow-up window and supports no general rule that traffic lights should be removed citywide.

A substantial part of the strategy was to demolish full cemeteries and rebuild them on the same site, which freed capacity in the game again. This is an operational intervention, not a lasting solution through added processing capacity. Work reports document that individual facilities were full again after a short time. The load is reset, but the underlying number of deaths and the road routes remain. Shortly before the end of the day, a complete warning inventory lists 129 messages, 16 of them funeral warnings. That is a favourable snapshot compared with the morning, though the ongoing interventions and the other warnings remain part of the balance.

Mixed use instead of further densification

The lower density was not only an autonomous decision. The work records hold a human request to investigate the causes of the supply problems and to review the concentration of high residential buildings. A pilot and a second stage, explicitly requested afterwards, replaced strips of high-rises with mixed-use buildings. Schools, offices and other facilities to be kept were deliberately left out.

The saved building queries of the four sub-areas checked later show 22 mixed-use buildings, so the conversion is more than an intention. Existing housing was lost as well, and projected later populations of the new building types must not be read as results already achieved.

Traffic observation was sobering. The follow-up report gives a congestion share of 59.6 percent before and 60.8 percent after in the north-east area; no improvement was visible in that comparison. A further reduction in density was therefore set aside for now, and a metro branch was built into the poorly served area instead. The drop in residents in the converted area does not count as an optimisation success in itself.

The effect on rail also remained open. The operating log documents added depot slots and more trains, but a later daily sample shows low use of the new line and continued heavy road congestion. The assumed mismatch between the line's route and actual destinations remains a hypothesis. These observations support neither a general ineffectiveness of the metro nor a demonstrated CPU advantage.

Mail: expansion confirmed, relief varies by site

In the evening two heavily loaded sorting facilities were extended with automatic sorting and additional loading bays. The saved building checks confirm active extensions at both and a fleet growing from five to ten vehicles each.

Stock has to be judged at matching comparison points. In the saved inspections of two consecutive rounds, the stock at the north-east facility fell and the west facility's rose slightly (exact values in the table). One work report uses a different, higher intermediate figure as the starting point for the west facility. These time windows must not be blended into one success curve. The expansion is documented; a backlog falling everywhere within the same window is not. Further observation has to show whether sorting and removal keep pace with intake over a longer time. The improved mail factor in a later city overview does not replace this local check.

Operational stability and limits on performance claims

Two unplanned game crashes with recovery from saved states are documented for the day, at about 04:00 and 14:03 UTC, followed by planned restarts. The building agent examined a possible link with growing memory use. The records support a working hypothesis, but neither a confirmed cause of the crashes nor a guaranteed effect of preventive restarts.

The qualified watchdog series of the regular performance file is missing again for 28 September, so the delivery marks the whole day as a measurement gap of unknown cause. That does not mean the game stood still: the many saved game states show operation, but they do not automatically belong to the qualified CPU series. No quantified performance gain follows from the report. Selected speed, actual simulation throughput and CPU effort per unit of simulation work must be considered separately, and a lower population after demolition must not be presented as saved computing work at equal city performance.

Status

The daily file lists 13 gap entries, and missing sections and uneven field coverage have to stay visible. Whether the calmer warning counts hold, and whether mail and traffic improve under comparable conditions, remains open. The report ends on 28 September at 22:00 UTC; later interventions belong to the next day. Project page: CS2.