What Happens When You Map 831 Russian Prosecutions to the Sustainable Development Goals

Political prosecution targets more than rights. It targets development.

When governments report progress on the UN Sustainable Development Goals, political prisoners do not appear in the data. Prosecutions are classified as criminal justice. The people they remove from society (researchers, journalists, activists, religious minorities, organizers) are classified as offenders. The development domains those people serve are never named. This analysis names them.

Russia has politically prosecuted 7,132 people in the OVD-Info dataset. Each prosecution is filed under a specific Criminal Code article. Each article targets a specific category of person. And each category of person corresponds to a specific SDG domain. Art. 280 (extremist speech) silences the academics and journalists who sustain SDG 4 (Quality Education). Art. 275 (treason) criminalizes the international scientific collaboration that sustains SDG 9 (Innovation). Art. 205.5 (terrorism) targets the ethnic and religious minorities whose inclusion defines SDG 10 (Reduced Inequalities). Art. 282.1 (extremist organization) liquidates the NGOs that sustain SDG 17 (Partnerships for the Goals).

We mapped every prosecution in the dataset to the SDG it most directly damages, using a SKOS vocabulary that links each Criminal Code article to its functional target. The result is not a correlation between prosecution and development decline. It is a direct count, validated against the 217,814-triple knowledge graph: 831 prosecutions classified as potentially impeding SDG 4. 1,950 filed against SDG 10. 238 filed against SDG 9. 83.6% of the SDG 4 prosecutions happened after the February 2022 invasion. 4,040 people face charges that damage two or more SDG domains simultaneously. 15 of those people are children.

The implication for policy: political prosecution is conventionally placed under SDG 16 (Peace, Justice and Strong Institutions) and treated as a governance problem. This analysis shows it is also an education problem (SDG 4), an innovation problem (SDG 9), an inequality problem (SDG 10), an economic growth problem (SDG 8), and a partnerships problem (SDG 17). Governments that prosecute political dissent are not failing on one SDG. They are failing on six. International development institutions (the World Bank, UNDP, bilateral aid agencies) that treat governance and development as separate policy domains are measuring the damage with the wrong instrument.

Which SDG does each prosecution category damage?

The invasion changed which SDGs are under attack.

How each prosecution category damages a specific SDG.

Three findings that required the knowledge graph.

The charge-to-SDG mapping can be computed in Python. The following three findings require connecting four layers of data simultaneously: the person (age, gender), the prosecution (charge, year), the Criminal Code article, and the SDG domain it damages. Any relational database or data analysis tool could produce the same results with the right joins. The knowledge graph makes these queries composable without schema redesign, reusable across jurisdictions, and extensible as new SDG mappings or country datasets are added. That is a practical advantage for a multi-country research program, not a computational requirement.

Finding 1: Children prosecuted per SDG domain

Finding 2: Compound SDG damage per person

Finding 3: Russia applies its harshest charges exclusively to men

288 people have been charged under Art. 205.3 (terror training), the charge that carries the longest sentences in the political prosecution system. All 288 are male. Zero are women. This is not a sampling artifact. It is a pattern: the Russian state applies its most severe political prosecution instrument exclusively to one gender.

We report this as a standalone prosecution finding, not as an SDG indicator. The standard UN SDG 5 framework focuses on empowering women and closing gender gaps. The 0/288 pattern is the opposite: men bear the heaviest prosecution burden. It is a gendered institutional pattern worth documenting, but it does not belong inside the SDG mapping framework. We initially classified it under SDG 5 and have moved it here on review.

Across all SDG domains, women comprise 17.5% of SDG 16 prosecutions, 22.2% of SDG 17, 10.7% of SDG 10, 16.3% of SDG 4, and 0.0% of SDG 5. The harshest charges (terrorism) are applied almost exclusively to men. The development consequence: the gendered prosecution pattern shapes which SDG domains absorb the most severe sentences.

This is not a correlation. It is a mapping.

Three kinds of claims exist in this space. This analysis makes the third.

Correlation runs a regression: prosecution count goes up, life expectancy goes down, a p-value falls below 0.05. The problem is that both variables are driven by the same political crisis. The regression finds that bad things happen together. That is not a finding. We ran this analysis (panel fixed-effects, four countries, 639 observations). It produced statistically significant but substantively ambiguous results, because prosecution and SDG decline are both symptoms of the same event.

Causation compares an observed trajectory to a counterfactual: what would Russia's SDG indicators have done if it had not escalated prosecution? This requires a synthetic control (Abadie et al., 2010) or a difference-in-differences design with a non-escalating comparator. We attempted this using 12 control countries. It returned no significant results because the post-escalation time series are too short (2-3 years) and the confounders (sanctions, war, mobilization) are too large. This analysis will gain power as more data accumulates. It is not ready.

Mapping does neither. It classifies each individual prosecution by the SDG domain it impedes, based on the legal function of the charge article. Art. 354.1 (rehabilitation of Nazism) is a speech crime applied to academics and students. Speech crimes against academics impede education. Education is SDG 4. Therefore this prosecution maps to SDG 4. The number 831 is not a regression coefficient. It is a count of prosecutions whose charge article, by its legal function, impedes SDG 4. No p-value is needed because no statistical inference is being made. The claim is taxonomic, not causal.

The strength of mapping is that no confounder can invalidate it. Sanctions did not cause 831 prosecutions to be filed under speech articles against people in educational contexts. The state did that. The mapping counts what the state did, classified by which development domain it targeted.

The weakness of mapping (identified through red-teaming) is that the taxonomy is analyst judgment. Art. 275 (treason) maps to SDG 9 (innovation) only if the defendant is a scientist. If the defendant is a military officer, the mapping is wrong. The OVD-Info dataset does not include defendant occupation, so the knowledge graph cannot disambiguate at the person level. The SDG 9 count (238) should be read as an upper bound. A stronger version of this analysis would filter treason cases to those where journalism or court records identify the defendant's profession.

How to read the numbers.

Double-counting rule: A single prosecution can map to multiple SDGs. Art. 280 (extremist speech) maps to both SDG 16 (institutions) and SDG 4 (education). The 5,376 SDG 16 prosecutions and the 831 SDG 4 prosecutions overlap. The total number of unique persons with SDG-mapped charges is 5,700. Because individual articles map to multiple SDGs (Art. 280 maps to both SDG 16 and SDG 4), the total count of person-SDG pairings is higher than the person count. Both numbers are correct. The first counts people. The second counts the development domains each person's prosecution damages. This report uses the SDG-event count (~11,100) because the argument is about breadth of damage, not headcount.

What the mapping measures and what it does not: The charge-to-SDG mapping is an analyst judgment, not an automated classification. Each link is defensible (Art. 275 treason against scientists \uc0\u8594 SDG 9 innovation is a direct causal mechanism), but alternative mappings are possible. The mapping does not measure the magnitude of damage per prosecution. A 20-year terrorism sentence removes more human capital than a 3-year speech charge. A future iteration should weight by sentence length.

SDG 13 (Climate Action): The Russian dataset contains few prosecutions of environmental activists specifically. But for the proposed cross-country expansion (Azerbaijan, Georgia, Egypt, Nicaragua, Venezuela), prosecution of environmental defenders maps directly to SDG 13. The framework includes SDG 13 as a placeholder for countries where that pattern is documented. This brings the total framework to 9 SDG domains, of which 8 are populated from the Russian data.