Public safety & policing

In-depth analysis

KB1 — Criminal-justice data platform

• Mechanism: The statistics of the current police (National Police Headquarters — ORFK), prosecution service (Prosecutor General’s Office) and prison service are migrated to a unified, public platform, disaggregated at district level. The platform visualises offence types, clearance rates, procedural duration, victim demographics and recidivism data. The key point: the data are not merely aggregated but available as time series (trend analysis) and comparable with police resource allocation (more officers = fewer crimes? — measurable).
• Quantified target: By 2028, the platform covers 90% of offence types at district level; the correlation between clearance rate and resource allocation is publicly analysable; data are updated quarterly.
• International precedent: United Kingdom — the parallel system of the Crime Survey for England and Wales (CSEW) and police recorded crime statistics is one of the EU’s best criminal-justice data infrastructures. The CSEW matters because police statistics measure reported offences, whereas the survey captures actual victim experiences — the gap between the two is the “dark figure”.
• Trade-off / risk: Publishing crime data at district level creates a “crime map” that can stigmatise settlements — property prices and investment appetite may fall. The solution: the data must be contextualised (trend, comparison, intervention effect), not published as raw numbers alone.

KB2 — Community policing

• Mechanism: The essence of the community policing model: the officer does not merely patrol and punish, but is a permanent, familiar member of the local community. In every disadvantaged (LHH) settlement above 5,000 inhabitants, one dedicated community officer who: (1) attends local municipal meetings; (2) is available weekly at a “surgery hour”; (3) cooperates with school social workers and family support services. The community officer is evaluated not by the number of penalties issued, but on the trust index (annual local survey) and crime-prevention indicators.
• Quantified target: By 2030, 200 community officers in disadvantaged settlements; trust in the police in target areas rises by 20 percentage points (current estimate: ~40% in segregated areas, national: ~60%); minor offences (theft, vandalism) fall by 15% in target areas.
• International precedent: Scotland — when Police Scotland was created in 2013, community policing was made the operating philosophy of the whole organisation, not merely a “special programme”. The result: police use of force fell by 40%, community trust grew, and crime continued to decline — all this in one of Europe’s previously most troubled regions (Glasgow).
• Trade-off / risk: A community officer’s effectiveness depends on personal aptitude — the model cannot be scaled by simply increasing headcount. Poor selection (an authoritarian personality in a community role) is counterproductive. Moreover, the “soft” outcomes of community policing (trust, prevention) are politically harder to communicate than “hard” indicators (arrests, penalties).

KB3 — Crime prevention vs. punishment impact assessment

• Mechanism: Control-group (RCT or quasi-experimental) impact assessment of the following question: does a given budget amount spent on crime prevention (education, employment, community-building, lighting, CPTED — Crime Prevention Through Environmental Design) or on punishment (more officers, harsher sanctions, prison) reduce crime more? The assessment runs in 10 selected districts over 5 years, with independent academic evaluation (HAS Centre for Social Sciences / National Institute of Criminology — OKRI). The results feed into budget allocation.
• Quantified target: The impact assessment starts by 2028, first results by 2033; the share of crime-prevention spending in the law-enforcement budget rises from 5% to 15% (if the results confirm effectiveness).
• International precedent: Washington State Institute for Public Policy (WSIPP) — regularly conducts cost-benefit analyses of criminal-justice interventions. Finding: 1 dollar spent on prevention yields on average 7–10 dollars of social savings (reduced prison costs, victim costs, lost productivity), while imprisonment on its own does not reduce recidivism.
• Trade-off / risk: The 5-year horizon of an impact assessment is politically “too long” — research that spans electoral cycles needs institutional protection. Moreover, the “prevention is more effective” finding is politically difficult to communicate, as the public intuitively views harsher punishment as more effective (the psychological basis of the “retributive urge” is strong).

KB4 — Reintegration

• Mechanism: The reintegration programme for released prisoners rests on three pillars: (1) In-prison preparation: intensive vocational training in the final 6 months (integrated with FO3), basic financial literacy, labour-market counselling. (2) Transitional housing: a “halfway house” system (12-month transitional accommodation with supervision and mentoring), which is neither prison nor independent housing. (3) Employment support: an employer hiring a released prisoner receives a subsidy equivalent to 50% of wage costs for 12 months. The programme’s effect is tracked at individual level (recidivism rate at 1, 3 and 5 years).
• Quantified target: Reducing the recidivism rate among programme participants from ~50% to 35% within 5 years; 2,000 released prisoners take part each year; halfway-house capacity of 500 places.
• International precedent: Norway — the world’s lowest recidivism rate (20%). The Norwegian prison system rests on the combination of the “normality principle” (prison resembles free life as closely as possible) and an extensive reintegration programme (Friomsorgen — post-release aftercare). In-prison training and work are not optional but expected. The cost per inmate is high (€90,000/year), but the social return (low recidivism) compensates.
• Trade-off / risk: The reintegration programme is costly (a halfway-house place costs ~€15,000/year), and the “we are helping criminals” narrative is politically vulnerable. The Norwegian model presupposes high social trust — in Hungary, where trust in the prison system is low and public punitiveness is high, communication is critical. The most effective argument is budgetary: reducing recidivism reduces prison costs.

KB5 — Predictive policing with an ethical framework

• Mechanism: Two types of predictive tool: (1) Spatial (hotspot policing): hotspots are identified on the basis of crime data — resource allocation concentrates on them (this is proven, and ethically less problematic). (2) Temporal (predictive timing): the temporal pattern of crimes is analysed (e.g. weekend nights, payday) and patrols are scheduled accordingly. What we do NOT do: individual-level prediction (person-based profiling) — this is excluded both ethically and under the EU AI Act. The code of every algorithm is public, with annual bias testing (results checked by ethnicity, age and sex). Oversight is carried out by the Ethics Council.
• Quantified target: Hotspot policing rolled out in the 20 largest cities by 2029; a 10% drop in crime in target areas; algorithmic bias testing annual and public; individual-level prediction excluded with zero tolerance.
• International precedent: Los Angeles (USA) — the PredPol (now Geolitica) hotspot system was introduced in 2012; the initial results were positive (−7% in burglary), but the system was shut down in 2020 after it emerged that bias in historical data (overpolicing in Black neighbourhoods) had created a self-fulfilling prophecy. Amsterdam, by contrast, built a successful model: it uses purely spatial prediction, without individual profiling, and audits the algorithm annually.
• Trade-off / risk: Even “purely spatial” prediction can be indirectly profiling: if the “hotspots” fall in ethnically segregated areas, the intensified police presence there amounts to de facto ethnic profiling. The solution: hotspot analysis must be restricted to breakdowns by offence type and time period, and the designation of a “hotspot” is supervised by the Ethics Council.

KB6 — Police code of ethics and accountability

• Mechanism: A public, auditable code of ethics for the police: the framework for lawful use of force and the protection of citizens’ rights. An independent complaints commission with civilian participation investigates abuses. Locke: the purpose of political power is the protection of rights, solely for the common good. Hobbes: the purpose of public power is to guarantee security.
• Quantified target: Code of ethics adopted by 2028; 100+ complaints-commission investigations per year; the number of police abuses falls by 30%. 📖 Source: Locke: Second Treatise of Government; Hobbes: Leviathan

KB7 — Anti-corruption internal oversight reform

• Mechanism: Kautilya: the supervisors too must be supervised, for “rules alone do not prevent unethical conduct”. Strengthening internal control at the National Police Headquarters (ORFK) and law-enforcement bodies: rotating internal audits, asset-declaration monitoring, informant-protection programme.
• Quantified target: A rotating internal-audit system 100% in place by 2029; 200+ internal investigations per year; the clearance rate of law-enforcement corruption cases rises to 50%. 📖 Source: Kautilya: Arthashastra

KB8 — Victim-support service minimum

• Mechanism: Locke: the state’s primary duty is the protection of citizens’ persons and property. A nationwide, uniform victim-support system: immediate legal, psychological and financial support for crime victims. “Victim-support points” in county-rank cities, online access in small settlements.
• Quantified target: 23 victim-support points (one in every county-rank city) by 2029; 10,000+ victims supported per year; victims contacted within 48 hours in 80%+ of cases. 📖 Source: Locke: Second Treatise of Government

KB9 — Crime Prevention Through Environmental Design (CPTED)

• Mechanism: Hobbes’s logic: where there is no “common power to keep them in awe”, order breaks down — reshaping the physical environment is also order-making. A municipal public-safety audit using the CPTED methodology: improving public-space lighting, visibility and natural surveillance in the areas with the highest crime rates.
• Quantified target: Public-safety audits in 50 settlements by 2030; a 15% drop in public-space crime in target areas; public-space lighting coverage of 90%+ in target areas. 📖 Source: Hobbes: Leviathan