AltTab

Tarkov's language gap is 48 points, and it is about servers

Escape from Tarkov News 6 min read

Data basis: eight per-language pulls from Steam’s public appreviews endpoint (appid 3932890), 100 reviews each, read 2026-08-27, most recent first, plus a deeper 200-review Simplified Chinese pull for topic analysis. Topic classification uses a Simplified Chinese keyword set. Quoted text is translated.

Escape from Tarkov by language Positive rate, 100 reviews per language, read 2026-08-27 Portuguese (BR) 79% English 67% French 63% German 61% Russian 60% Korean 52% Japanese 51% Chinese (Simpl.) 31% A 48-point spread. Russian sits mid-table despite the studio being Russian.

31% against 67%

Escape from Tarkov’s English reviews run 67% positive. The Simplified Chinese pull on the same day runs 31% — a 36-point gap, and 48 points from top to bottom once Brazilian Portuguese (79%) is included.

One result cuts against the obvious guess. Russian sits mid-table at 60%, close to German and French, despite Battlestate Games being a Russian studio. Whatever is driving the Chinese figure is not simply distance from the developer.

One topic carries it

What the Chinese-language negatives are about Tarkov, 139 negative reviews from a 200-review Simplified Chinese pull Servers / performance 27% (37 of 139) Seasons / updates 5% (7 of 139) Price / monetisation 3% (4 of 139) Chinese keyword set. One topic dominates; the rest barely register.

Of 139 negative Chinese reviews, 27% mention servers, optimisation, lag or crashes. Nothing else comes close: seasons and updates appear in 5%, price in 3%.

The reviews are specific about which servers. One 214-hour reviewer writes about paying for a season pass and getting no optimisation on the regional servers in return. The complaint is not that the game is badly made — many of these reviewers have hundreds of hours — but that the infrastructure they connect to does not match what they paid for.

That is a different complaint from the English corpus, where performance leads at 22% but cheating follows at 14% and monetisation appears at 8% against 1% in positives. In the Chinese pull, monetisation on its own barely registers; it shows up welded to the server complaint.

What this changes

  • If you read the store page in Chinese, Steam’s own language filter shows you 31%. The headline figure is not yours.
  • If you are choosing a region to play in, this is the clearest signal in the dataset — the gap tracks infrastructure, not translation or content.
  • If you are reading Tarkov’s score generally, the six-game language map shows Tarkov has the second-widest spread of any game measured here. A single percentage is doing a lot of hiding.

Method, and its limits

Rates are positive reviews divided by total, per language pull — language-neutral and reproducible. Topic classification is keyword matching with a Simplified Chinese keyword set; one review can match several topics, and matching is blind to sentiment inside the review.

No topic analysis is reported for the other six languages, because a validated keyword set does not exist for them here. Running English keywords over Japanese or Portuguese text produces near-zero matches, which would read as “these players do not complain about servers” when it actually means the method did not look.

What could not be confirmed

  • Whether the gap predates the Steam release or Patch 1.1.0.0. The endpoint returns recent reviews only.
  • Which specific servers or regions the complaints refer to. The reviews name them colloquially; this page does not resolve those names to infrastructure.
  • Whether language equals region. A Simplified Chinese review can be written anywhere.
  • Cheating rates by language. The Chinese keyword set includes cheating terms; they did not clear the reporting threshold, which is not the same as their being absent.

Sources

Related posts