A fast answer to the only question that matters. We judge it by the live Steam player count: if people are on the servers and the data is fresh, everything works.
First day of the playtest: roughly two-hour queues for the servers, with the developers running a live infrastructure ticker.
BULKHEAD has no official server-status page — not on the game site, not on Steam. So we measure it indirectly, but with our own hands: we poll the public Steam API every 10 minutes and record how many people are in the game. If the last sample is fresh and players are in, the servers are up.
The three states you see above mean this:
| Up | sample newer than 40 minutes, players in the game |
| Empty | sample is fresh but the count is zero — the usual picture when servers are off between tests |
| No data | the last sample is stale: our collector fell over, or the Steam API is not answering |
This method honestly catches global outages and shutdowns. What it will not see: a single-region failure, broken matchmaking on live servers, or a problem with one particular ISP. If the count is high and you still cannot get in, the problem is almost certainly local to you.
We do not copy anyone else's figures: the samples are our own, every ten minutes, through the public Steam API. That is why our chart is almost always slightly different from SteamDB or any other tracker — they poll at their own moments and average in their own way. A few percent of difference is normal, and it means exactly this: both sides measure honestly, just not in the same second.
The second difference matters more: we show the gaps. If the collector was down, the chart has a visible break labelled «no samples» rather than a straight line drawn across the night. A straight line there would be an invention, and inventing data in a project that lives on measurement is the worst thing you can do.
Since 25 August the sampling runs on our own cloud service rather than a home computer: it goes around the clock and does not depend on anyone leaving a machine switched on.
Right now every number on the site comes from the playtest — a separate Steam app. The main game is not out, and Steam simply has no player counter for it: the request returns an error. So wherever you see a player count, it is the test build being counted.
When Early Access launches, the main game's counter starts working and we begin writing both series: the release and whatever is left of the tests. Nothing needs switching — the charts pick up the new data by themselves, and the playtest history stays for comparison.
Then the servers are alive and the problem is on your side or your ISP's. In order: restart Steam, verify the game files, turn off your VPN, reboot the router.
If none of that helps, write in the chat (it is in Ukrainian): if several people see the same thing at once, it is no longer a local problem and we raise the alarm on the channel.
Because Steam hands out the player count in slices, not as a stream. Polling more often gains nothing — the number does not get more accurate — while the risk of being rate-limited grows. Ten minutes is the compromise that keeps the chart alive and the data honest.
Same source — the public Steam API — but the samples are ours, and we have kept the history since 24 August 2026, before launch. A discrepancy of a hundred or two players between services is normal: it all depends on which second the slice was taken.
A profile keeps your contracts and listings with you: sign in from your phone or your PC and everything is still there. You can post without one, but then the records live only in this browser.