What the torture is for

Two states, one tool, two functions. In Belarus the documented violence sits at the point of capture. The beating happens in the street, in the police van, at the checkpoint — 32.1% of documented cases where the person was taken at a protest carry a torture or beating report, and 25.6% where they were stopped at the border, against 7.4% where the person was picked up later, by name, once the case already existed. That shape is punitive: by the time a Belarusian investigator opens the file, the evidence — a post, a subscription, a transfer — already exists on a server, and the state needs nothing further from the suspect's body. The violence is the message, and it is delivered in public. Russia's shape is the inverse of this. Its street arrests carry almost no torture reports; the violence appears after the door closes, and it clusters with surgical precision in the cases whose evidence only the suspect can produce: 20.5% of device-evidence cases carry a torture report and 16.2% of money-trail cases, against 5.2% of public-post cases — a 3.9× gradient inside the same court system. The case files say what the beatings were for: a pawnshop cashier "tortured into confessing," an artist tortured "with electricity and water," a programmer beaten and "pressured to act as an informant." Belarusian torture is a punishment. Russian torture is procurement — of passwords, signatures and informants — and the gradient is the receipt.

Bases with at least 25 documented Russian cases. Red = the two evidence channels that must be produced out of the suspect himself — a device unlocked, a transfer explained. The gradient runs opposite to visibility: the loud public cases carry the fewest torture reports.

Arrest contexts with at least 20 documented Belarusian cases. Red = capture in the open — a protest, a border stop. Violence concentrates at the moment of capture and falls away once the case moves indoors: the mirror image of Russia.

• §torture_reported records that a source documented torture or beatings in that case — absence of a report is not absence of torture. Several countries in our database sit at zero purely because their source networks do not record it. That is why this finding never compares torture LEVELS across countries — only the structure of where reports fall within each country's own documented cases.
• §Documentation practice shapes the structure too. Belarus's post-2020 street beatings were exhaustively documented at the point of detention; Russian pre-trial torture surfaces mainly through defense lawyers in closed cases. Both patterns are read from what each country's monitors were able to record.
• §Some cells are small: the Belarusian protest figure rests on 28 documented cases, the Russian device-evidence figure on 39. Every bar carries its case count, and every underlying case is in the receipts.
• §The basis and arrest-context of each case are editorial classifications drawn from cited evidence, each with a confidence score and a verbatim quote (shown in the receipts).
• §These are cases documented in our database (primarily from the Viasna Human Rights Centre), not a count of every prosecution. The value here is case-level detail: what each person actually did, with a source. Russian cases are heavily sourced from Memorial, OVD-Info and Mediazona.

Every documented case behind this finding, each linking to its profile and primary source. 122 carry a verbatim quote of what the prosecution rested on.