A single hosting region is usually explained as a cost decision, and it is one, but it is worth being precise about what is being bought. The saving is server capacity in additional locations. The cost is the portion of the audience that cannot play the game well, and this log is an attempt to size that cost.
The shape
All six single-region games in the archive follow the same curve. Month one looks normal, because launch interest overwhelms everything. By month three the far-continent population is falling faster than the host continent's. By month twelve the game has settled at roughly a third of its opening figure, and almost everyone remaining is near the server.
| Month | Single-region median | Multi-region control median | Gap |
|---|---|---|---|
| 1 | 100% | 100% | — |
| 3 | 54% | 76% | 22 pts |
| 6 | 39% | 67% | 28 pts |
| 12 | 31% | 58% | 27 pts |
| 24 | 28% | 51% | 23 pts |
Note that the gap opens early and then stops widening. This is what makes the finding usable: the damage is done in the first six months and it is essentially permanent, because the far-continent players who left are not waiting for a second region to be announced.
Where the losses sit
Seventy-nine per cent of the decline relative to controls comes from continents distant from the host. Sixteen per cent from the adjacent continent. Five per cent from the host continent itself, which is within the noise of ordinary attrition.
In other words, the game does not become less appealing. It becomes unavailable to a specific set of people, in a way that neither they nor the studio ever describes in those terms. Our Frost Relay review is the clearest current example: excellent design, one region, and a review score two points below what the same game would earn hosted properly.
Nobody writes a review saying the netcode drove them away. They say the game felt bad, and they are right, and the reason is a decision made in a spreadsheet.
What cooperation does to the numbers
The effect is strongest in cooperative games and weakest in games with asynchronous or turn-based interaction, which is intuitive but worth quantifying. Of our six single-region titles, the two with real-time cooperation settled at twenty-four and twenty-seven per cent. The one with primarily asynchronous play settled at forty-eight.
Latency destroys shared action before it destroys anything else. Handing an object to another player, occupying the same doorway, or repairing the same panel all require agreement about the present moment, and that is the first thing a quarter of a second removes.
The mitigation nobody uses
None of the six games shows the player their round-trip time, and none has a region selector. That means a distant player has no way to understand their own experience and no way to make an informed decision about whether to keep trying.
Slipstream Nine shows estimated latency to all five of its regions before you queue. It cannot manufacture a nearby server, but it can stop a player from concluding that the game is badly made. Of the twenty games in this log's sample, three do this. All three are in the top quartile of our review scores.
Findings
- Single-region games settle at a median thirty-one per cent of month-one population by month twelve.
- Multi-region controls settle at fifty-eight per cent over the same window.
- Seventy-nine per cent of the relative decline comes from distant continents.
- The gap opens by month six and does not widen afterwards.
- Real-time cooperative games suffer roughly twice the loss of asynchronous ones.
- None of the six single-region games discloses latency or offers a region selector.
Status
Published, with a caveat we want stated plainly: six games is a small sample for a population claim, and observed player counts are noisier than our packet measurements. The finding survives removing any one game, but we would not defend the precise thirty-one per cent figure to better than five points either way.