We ship an onboarding system called First Shift. It runs only from the UI open event on a wall-mounted noticeboard entity, SolreignWingmateBeacon. There is no spawn hook. If a player never opens that board, First Shift never runs for them.
We have seven rotation maps: Terminus, Leviathan, Perihelion, Verdant, Oasis, Meridian, and Nocturne. On each map the beacon sits somewhere relative to the latejoin spawn cluster. We measured Chebyshev tile distance from the map files, not by eye:
| Map | Tiles to beacon | |---|---| | Nocturne | 6 | | Meridian | 7 | | Terminus | 31 | | Leviathan | 35 | | Perihelion | 35 | | Verdant | 35 | | Oasis | 59 |
On five maps the distance is 31 to 59 tiles. On two it is 6 or 7. Meridian was previously treated internally as the layout to copy; at 7 tiles it is the second-closest, not zero.
Large spawn-to-beacon distance is a map placement issue, not a defect in the First Shift code. The code runs when the board is opened. Whether the distance reduces how often players open it is unmeasured, and the next section is about why.
The fix is a map coordinate edit on the existing grid parent. Each beacon already shares its map's grid parent with that map's arrivals marker, so moving it needs no code change and no new entity. The engineering lane has scoped the work and approved it in principle.
This fix is not shipped. It is not committed to any branch. The current live server build predates it. Nothing in production has moved.
Coordinate moves are cheap. What we still lack is session telemetry. We cannot measure how many players opened First Shift. The server does not log ready-state, spawn-versus-no-spawn, or whether the low-population lobby reminder was delivered. Every previous conclusion we reached about an arriving player's session was reconstructed from console commands that happen to be logged. Until that changes, the placement move is a hypothesis we can implement, not a claim we can verify after deploy.
What we do have is a public arrival log. It posts every connection and departure to a Discord channel with no usernames. Between 2026-07-30 and 2026-08-02 it recorded eleven sessions. Their lengths: 2m, 2m, 6m, 6m, 8m, 16m, 16m, 22m, 28m, 50m, 366m.
On six of those sessions the server posted that a round started after somebody waited in the lobby. Five of those six had exactly one player aboard. A single player can start a round.
One earlier session is known in more detail. The player connected, stayed about seventeen minutes, issued a ready toggle and then a restart vote, and never spawned. Late joining was never blocked for them. We do not know why they did not get in. Console commands tell us the sequence. They do not tell us intent, UI state, or whether First Shift was ever presented. A round starting is not the same event as a player spawning, and that session is the proof.
So the picture we can defend is narrow. First Shift only runs from a board open. On five of seven maps that board is 31 to 59 tiles from latejoin spawns, measured from the map files. Moving the beacons is scoped and approved in principle, and it is not live. The arrival log includes session lengths from 2m to 366m, six lobby-wait round starts, and one connect that never became a spawn. It does not show open rates for First Shift, and we should not pretend it does.
Next: commit the beacon moves, and add session telemetry that records First Shift opens, ready state, and spawn.