Method: how this site decides
There is no official ROB RUN server API. Every status site is therefore inferring. This page writes down exactly how the inference works, so you can judge it instead of trusting a green dot.
The Steam Web API exposes a single number for the whole game: players online right now. Steam's News API exposes developer announcements. The third input is what visitors report on this site. Nothing else is public — not per-server health, not queue depth, not regional latency.
A Cloudflare cron reads the player count every five minutes and stores it with the exact observation time. Failures are stored as unavailable, never as zero. A reading older than thirty minutes is not used to claim anything about the present.
The baseline is the mean of successful samples taken 60 to 120 minutes before the latest reading. A drop of 35% or more is published as degraded; 60% or more is published as a suspected outage. A zero reading is published as an unconfirmed outage, because Steam occasionally returns zero while rebuilding its cache.
Anonymous reports are counted in fifteen-minute windows. More than 25 in a window, or three times the three-day average, pushes the verdict to at least degraded. The note text is never published verbatim, and the IP is only hashed for rate limiting.
A single good reading does not clear an incident. The board returns to normal only after three consecutive calm samples — within 20% of baseline — with reports back near their usual rate. This is deliberate hysteresis: it stops a flapping counter from producing fake all-clears.
- A healthy count never means every server is healthy. It means people are playing.
- A failed API call is never drawn as zero players on a chart.
- We do not publish per-server wipe times, because no public source exists.
- We do not invent incidents. If the developer has not spoken and numbers look normal, the page says normal.
| Signal | Threshold | Published state |
|---|---|---|
| Player count | ≥ 60% below 1 h baseline | Outage suspected |
| Player count | ≥ 35% below 1 h baseline | Degraded |
| Player count | Exactly zero | Unconfirmed outage |
| Reports | > 25 in 15 min, or > 3× 3-day average | Degraded at minimum |
| Recovery | 3 consecutive samples within 20% of baseline | Operational |
| Data | No successful sample for 30 min | Unknown — not offline |
Samples are kept indefinitely at five-minute resolution so long windows stay honest. Anonymous reports are kept for 30 days, then deleted. If a reading is later found to be wrong, the correction is written into the incident note rather than silently edited.