Sitting in Strasbourg since 1959, enforcing the Convention for the 46 Council of Europe states. Individuals petition it directly, making it by far the most active human rights court in the world - thousands of judgments a year. Its limit: regional, and dependent on state compliance (Hirst v UK went unremedied for a decade).
Sitting in Strasbourg since 1959, the court enforces the European Convention on Human Rights for the 46 Council of Europe states and delivers thousands of judgments a year - more activity than the rest of the rights system combined. The Council of Europe expelled Russia in 2022.
Use it as the rights half's institutional summit: a court that actually sits, rules and is obeyed, most of the time. Set it against the Security Council below: the Council of Europe expelled Russia in 2022, while the Security Council still seats Russia with a veto.
Individuals can petition the court directly and its judgments bind the 46 member states - a combination no other international court offers at this scale. Every EU member state is also a Convention state, so the rights order runs through the continent's deepest club as well.
Contrast the ICC below: binding judgments with individual access against a court that cannot arrest anyone.
Hirst v UK (2005), the prisoner-voting case, went unremedied for a decade - proof that even the strongest rights court depends on member-state compliance in the end. Abu Qatada v UK (2012) shows the other side: a state held to the torture prohibition even where it badly wanted otherwise.
The AO2 point: enforcement rests on state consent here just as everywhere else, but consent in Europe runs deeper than anywhere else.
A working rights order across 46 states, built case by case since 1959 - the rights regime's clearest delivered outcome. Compare the security half of the grid, where the defining recent record is Syria's half a million dead under a vetoed Council.
The qualifier: the order is regional. The court's success measures what consent-based law can do where consent exists, not what the global regime achieves.
Direct individual petition is the design feature that matters: a single prisoner (Hirst) or a deportee facing torture (Abu Qatada) can bring the United Kingdom to court and win. No security institution on the bottom half gives an individual any standing at all.
In an essay, this is the rights half's strongest claim to serve the least powerful - protection that does not wait for a state to act.
The court's reach stops at Europe's edge: the United States, China and the post-2022 expelled Russia sit outside it entirely. Its strength is bought by its selectivity - 46 broadly consenting states, not the world.
That is the same trade NATO makes on the UN-NATO grid: depth through like-mindedness, at the price of leaving the hardest cases untouched.
The strongest enforcement in the rights system, and the proof that rights protection works best where it is regional and voluntary. Thousands of judgments a year against the ICC's handful of convictions makes the point numerically.
The comparative line for the grid: the rights half peaks where consent is deepest; the security half fails where consent is needed most.
The permanent court for genocide, war crimes and crimes against humanity - with no police force, dependent on state cooperation for arrests, and operating without the US, China or Russia as members. Its docket has leaned heavily on weaker states, the standing selectivity criticism.
Created by the 1998 Rome Statute and open since 2002 in The Hague, the ICC is the first permanent court for genocide, crimes against humanity, war crimes and aggression - the institutional completion of the arc that began with the 1948 Genocide Convention and ran through the ad-hoc ICTY (1993) and ICTR (1994) tribunals.
As architecture it is the regime's summit; the columns to the right measure what the architecture can actually do.
The court has no police force: arrests depend entirely on state cooperation, and warrants against Omar al-Bashir of Sudan went unenforced for years while he travelled. The March 2023 warrant for Vladimir Putin - the first against a sitting head of a P5 state - measures how far the court route has travelled and how exposed its enforcement gap remains.
Compare the Security Council below: Chapter VII powers the ICC can only envy, frozen by a veto the ICC does not have.
Bashir travelled freely for years under an ICC warrant, including to member states that declined to arrest him. A court whose indictees can tour the world is a court whose compliance column reads minus.
The AO2 explanation: sovereignty is the wall - courts cannot reach defendants the host state will not surrender, the same wall the security half hits at the veto.
Convictions are few and slow: Thomas Lubanga (DRC, 2012) was the first, with Jean-Pierre Bemba and Dominic Ongwen following, and the Lubanga trial alone ran from 2006 to 2012. Against atrocities in progress, a trial cannot keep a village from being burned this afternoon.
The defence for AO3: the court's real product is the norm that sovereign immunity does not cover atrocity - slow justice, but a standing threat no leader faced before 2002.
The selectivity charge: the court's docket has leaned heavily on weaker states - its major convictions (Lubanga, Bemba, Ongwen) are all African - and the African Union has at times discouraged cooperation with it. The strongest powers prosecute no one and cannot be prosecuted.
Balance it with the Putin warrant of 2023: the docket is widening, even if enforcement is not.
The United States, China, Russia and India never joined; the court has 124 state parties and the gap is precisely the states whose conduct matters most. The enforcement gap was built in from the start of the Rome Statute system.
The realist point for the grid: a court the powerful opted out of mirrors a Council the powerful can veto - the same sovereignty problem in two institutional forms.
The permanent court the regime spent fifty years building cannot arrest, has convicted few, and operates without the US, China or Russia - yet it issued a warrant for a sitting P5 head of state in 2023, which no body on the security half has ever matched.
The essay line: the ICC is the rights regime's reach exceeding its grasp - judge it as norm-builder and it scores; judge it as protection and it does not.
The 1948 Universal Declaration and the 1966 Covenants (ICCPR, ICESCR) wrote the standards almost every state has signed. The regime's first arc - writing rights down - succeeded completely; the second arc, enforcing them, is the whole problem.
The 1948 Universal Declaration, drafted under Eleanor Roosevelt, plus the two 1966 Covenants - the ICCPR and ICESCR - form the International Bill of Rights, with treaty bodies and reporting cycles attached. The writing arc is almost complete: there is a treaty for every major right.
That completeness is the regime's first achievement; the rest of this row measures the second arc, enforcement. The security half inverts the pattern: the 1945 Charter wrote far less and armed it far more - Chapter VII powers from the start.
The UDHR is a declaration - morally weighty, not legally binding - and even the binding 1966 Covenants enforce through reporting and review, with no sanction behind them. The machinery names and shames; it cannot compel.
The grid's contrast: even the paralysed Security Council holds Chapter VII powers; the treaty system holds none at all.
Signature is cheap and nearly universal; practice varies wildly - states ratify the Covenants and violate them in the same decade, because the cost of signing is zero and the reporting cycle carries no penalty.
The AO2 explanation matches the climate regime's NDC problem: voluntary commitments produce voluntary results wherever interests cut against the text.
The normative victory is real and complete: every state now argues in rights language, the three generations of rights run from the 1948 UDHR through the 1966 Covenants to the 2010 UN recognition of the right to water, and sovereign immunity has been narrowed to the point where a P5 head of state carries an ICC warrant.
In an essay, this is the long-view case for the court-and-treaty route: it built the regime within which everything else, including intervention, now operates.
The standards exist precisely for people with no other protection - universal, inalienable and indivisible claims that any person holds against any power, a radical position barely older than the UN itself. Universal jurisdiction even let a Spanish warrant reach Pinochet in the UK in 1998.
The qualifier: a standard without a court is a promise, and most of the world's population lives outside any court that can enforce one.
Every major power signed the core instruments, and every major power applies them selectively - the same states that drafted the UDHR vetoed Syria resolutions and declined ICC membership. Endorsement of the words has never bound the conduct.
For the comparison: the security half at least admits great-power control openly (the veto); the rights half hides it behind universal signatures.
Standards without sanctions: the writing arc succeeded completely and the enforcing arc is half-built, which is the rights regime's whole story in one row. The gap between the 1948 text and the 1993 first tribunal - forty-five years - measures how long writing ran ahead of enforcement.
The judgement line: this row is the rights half's foundation and its ceiling - everything rests on it and nothing in it compels.
The only body that can authorise force - and the body the P5 veto can freeze. Syria is the defining failure: repeated chemical weapons use (Ghouta 2013, Khan Shaykhun 2017) met repeated Russian and Chinese vetoes, and the Council watched a civil war it could not touch.
The Council is the only body in the world that can lawfully authorise force, the centre of the system founded by the 1945 UN Charter with sovereign equality as its first principle. When it works - Resolution 1973 on Libya in 2011 authorised all necessary measures - nothing else in global politics matches it.
The grid's structural irony: the security half's strongest institution produces its weakest record.
Chapter VII powers are the hardest tool in international politics: binding resolutions, sanctions, and authorised force - the powers behind Resolution 1973 and the ICTY and ICTR tribunals the Council itself created in 1993 and 1994.
The teeth exist; the veto decides whether they are ever bared. That conditional is the whole security half of this grid.
Resolutions bind in law and are defied in practice when great-power protection is available: Syria ignored the Council for a decade because Russian and Chinese vetoes guaranteed no consequence would follow.
The comparison with the rights half: ECtHR judgments are mostly obeyed because compliance is the price of the European club; Council resolutions are defied because defiance can be vetoed into safety.
Syria is the defining failure: repeated chemical weapons use against civilians - Ghouta 2013, Khan Shaykhun 2017 - met repeated Russian and Chinese vetoes, and the war has cost over half a million lives with no intervention at scale. The Council watched a war it could not lawfully touch.
Set it against Rwanda 1994, where the peacekeeping force was reduced rather than reinforced as 800,000 died - the Council's failures recur.
Civilian protection depends on P5 politics: Benghazi was protected in 2011 because Russia and China abstained; Syrian civilians went unprotected because they did not. The same people the Charter promises to protect are hostage to the votes of five capitals.
For AO3, pair this with the ECtHR's individual petition above - the rights half gives individuals standing, the security half gives them odds.
The veto is great-power control written into the Charter by design - the price of getting the great powers of 1945 inside the system at all. Russia and China together vetoed multiple Syria resolutions; the Council has authorised no R2P-style force since Libya 2011.
The realist reading: the Council does not fail to manage great-power conflict, it was never built to - it manages everything except its own members.
Strongest powers, most predictable paralysis: the body that authorised Libya in March 2011 could not pass a Syria resolution for the rest of the decade. The formal powers and the actual record point in opposite directions, and the veto is the hinge.
The grid's judgement line: security fails exactly where it is global and vetoed, as rights succeed exactly where they are regional and consented.
Tens of thousands of blue helmets keeping ceasefires that would otherwise collapse - with no army of its own, troops borrowed from members, and a record that includes both quiet successes and catastrophic failures where mandates and means did not match.
A standing system of missions across continents, decades old, keeping ceasefires that would otherwise collapse - real machinery, even though every soldier in it is borrowed. It is the UN's most visible standing contribution to security.
For the comparison: it is the security half's nearest equivalent to the ECtHR - the workhorse institution that quietly delivers while the headline bodies fight.
No army of its own, troops borrowed from members, and restrictive mandates: Srebrenica in July 1995 is the standing warning - more than 8,000 Bosniak men and boys killed in a town designated a UN safe area, under UN watch, because the mandate and the means did not match.
The AO2 point: peacekeeping works by consent, which means it works exactly where the parties already want it to.
Parties mostly tolerate the blue helmets they invited - consent is the operating principle, and it holds because peacekeeping polices agreements the parties already made rather than imposing ones they did not.
Contrast R2P below: the moment force operates without consent, the regime needs the Council, and the Council needs the P5.
Ceasefires held for decades that would otherwise have collapsed - a genuine, sustained record, with the catastrophic exceptions (Rwanda 1994, where the force was cut as 800,000 died in a hundred days; Srebrenica 1995) marking where mandates and means diverged.
For AO3: judged against its own modest tools, peacekeeping is the security half's quiet success; judged against the atrocities it could not stop, it shares the half's failure.
The missions exist for the people the war is happening to - the only part of the security architecture physically present where civilians are at risk. The African Union's AMISOM mission in Somalia shows the model extending to regional bodies where the UN is stretched.
The qualifier is Rwanda: presence without mandate or means protected no one, and the force was reduced rather than reinforced.
Funded and mandated at the Council's pleasure: missions exist only where the P5 agree to create them and shrink when the P5 lose interest - Rwanda's force was cut in April 1994 at the moment it was needed most.
The structural point: even the security half's success story runs on great-power consent, which is the grid's recurring answer to every column.
Modest tools, genuine record: the contrast between peacekeeping's held ceasefires and the Council's Syria paralysis shows the security half works best at its least coercive. Consent-based operations succeed for the same reason consent-based rights courts do.
That symmetry is the grid's deepest finding: on both halves, the machinery that asks least of sovereignty delivers most.
Libya 2011 was the doctrine's high-water mark: Resolution 1973 authorised all necessary measures and NATO acted - then regime change went beyond protecting Benghazi and undermined the doctrine it invoked. Syria, vetoed and unprotected, is what R2P looks like when great powers say no.
R2P is a doctrine, not a body: coined by the Canadian-backed ICISS commission in 2001 and endorsed by every UN member state at the 2005 World Summit, it owns no machinery and borrows the Security Council's when the P5 allow.
That dependence is the design: R2P modifies sovereignty rather than replacing it - sovereignty is the default, intervention the exception.
Libya, March 2011: Resolution 1973 authorised all necessary measures to protect civilians in Benghazi - the first explicit pillar-three invocation - and NATO's air campaign began the same month. For one moment the doctrine had the hardest teeth in the system.
The contrast with the treaty system above could not be sharper: rights law that never compels against a doctrine that compelled exactly once.
One invocation, then the door closed: the operation expanded into regime change in the eyes of Russia and China, Gaddafi was killed in October 2011, and the Security Council has not authorised an R2P-style military action since.
The AO2 point: a doctrine that depends on great-power trust spent that trust in its first use.
Benghazi was protected, but the post-intervention Libyan state collapsed - and the backlash arrived exactly when Syria needed the doctrine: repeated chemical attacks (Ghouta 2013, Khan Shaykhun 2017), repeated vetoes, over half a million dead.
The line for the essay: R2P changed the language without changing the outcomes in the hardest cases - real norm, dead letter where great powers disagree.
The doctrine was written for civilians facing mass atrocity, and Syria's civilians - the precise case it was designed for - went unprotected through a decade of vetoes. Rwanda 1994, the moral fact that produced R2P in the first place, was effectively repeated under the doctrine's own watch.
For AO3: judged by its founding purpose, this is the security half's hardest failing.
R2P lives and dies by P5 consent: Russia and China abstained on Libya in 2011 and have refused R2P-style force ever since, vetoing Syria resolutions repeatedly. Kosovo 1999 showed the alternative - NATO acting without authorisation, judged illegal but legitimate - and the legal cost it carries.
The doctrine's career is the realist thesis in miniature: norms operate inside great-power consent, never above it.
Used once, burned once: endorsed by every UN member in 2005, invoked at full force in 2011, and effectively retired by the Libyan aftermath just as Syria presented the case it existed for.
The grid's closing comparison: the rights half built slow norms that survive (the Putin warrant); the security half built a fast norm that did not survive its first success.