A Complexity Scientist's View of Antarctica

What attention, decisions, and apparent agreement reveal about governing a continent without a ruler.

The problems we face increasingly cross borders, while the political institutions responsible for addressing them remain largely national. The United Nations has warned that exceeding 1.5°C of global warming is now all but unavoidable2. Scientists also report that humanity has crossed seven of the nine planetary boundaries that define a safe operating space for our planet3. Despite decades of scientific warnings and political debate, we still struggle to act together. Within a country, a government can decide and then enforce its decision. Between countries, no government can do either, so states have to agree before they can act together.

In one place, they have done so for more than six decades. Every year, twenty-nine countries meet to make decisions about Antarctica, and some of them are great powers that agree on little else. They decide by consensus: a proposal passes only when none of them objects, so any one of them can block it19. I would expect such meetings to deadlock. Still, they have maintained shared rules for human activity on the continent since the Cold War42,43.

I came to this puzzle as a complexity scientist. Complexity science is a fuzzy term, and I must admit the field comes with its own bullshit bingo card (emergence, self-organisation, networks). What I think it offers, like any study of systems that change over time, is a habit of mind. I ask how the parts of a system interact, and I study those interactions to see whether they explain how the whole system behaves. Here the parts are the countries and organisations at the table.

For the past year, I have studied how their cooperation endures and what happens when new problems need new rules. The meetings leave a long paper trail, more than sixty years of papers, decisions and reports. In it I look at three things: what the participants pay attention to28, how their rules build on earlier rules29, and how much support lies behind an agreement30. Together, they let me ask whether the practices that keep an institution going also limit its capacity to change31.

I wrote this essay partly to synthesise that research, and partly to introduce the Antarctic Treaty System, the set of agreements that governs the continent, to computational scientists. The topic may seem niche, but I think much of it is still unexplored. The system governs one of Earth’s largest global commons, and the setting is narrow and well defined enough to study how the actors interact.

A Crowded Continent

On a familiar world map, Antarctica is a thin strip along the bottom. Many of us picture ice sheets, penguins and a large, empty place far from our own lives. Viewed from above the pole, it is a continent where science, economic activity and strategic interests meet.

Antarctica: a place where interests meetA south-polar map highlights facilities, reported landing sites, and illustrative polar satellite paths as the adjacent text moves from science to economy to strategy.Belgrano II Antartic BaseCámara Antartic BaseCarlini Antartic BaseDecepcion Antartic BaseEsperanza Antartic BaseMarambio Antartic BaseMatienzo Antartic BaseOrcadas Antartic BasePetrel Antartic BasePrimavera Antartic BaseSan Martin Antartic BaseCaseyDavisEdgeworth DavidLawMawsonWilkins AerodromePrincess Elisabeth AntarcticaGoeldiComandante FerrazCriosfera 1Criosfera 2CrulsSt. Kliment OhridskiArturo Prat Antarctic Naval BaseBahía Fildes Maritime StationDr. Guillermo MannEstacion Polar Cientifica Conjunta Glaciar UnionGabriel Gonzalez VidelaLieutenant Rodolfo Marsh Martin AerodromeO'Higgins BasePresident Eduardo Frei Antarctic BaseProfessor Julio EscuderoRada CovadongaRipamontiRisopatronYelchoGreat WallKunlunQinLingTaishanZhongshanZhongshan SkiwayJohann Gregor Mendel Czech Antarctic StationPedro Vicente MaldonadoRepublica del EcuadorD85 Skiway campDumont d'UrvilleRobert GuillardGerman Antarctic Receiving Station (GARS)GondwanaKohnenNeumayer IIIBharatiMaitriBrowning PassConcordiaEnigma LakeMario ZucchelliMid PointTalos DomeS17 CampSyowaDirck Gerritsz LaboratoryScott BaseTorTroll StationMachu PicchuHenryk ArctowskiMountain EveningJang BogoKing SejongBellingshausenBunger OasisDruzhnaya IVLeningradskayaMirnyMolodezhnayaNovolazarevskayaProgressRusskayaVostokSANAE IVSANAP Summer StationGabriel de Castilla StationInternational Field Camp Peninsula ByersJuan Carlos ITurkish Scientific Research CampVernadskyFossil BluffHalley VIRotheraSignySky BluAmundsen-Scott South PoleMarble Point Refueling FacilityMcMurdoPalmerSiple DomeWAIS DivideArtigasRuperto Elichiribehety
Facilities listed by COMNAP, south of 60°S.The coastline is from Natural Earth, and landing sites are from treaty information-exchange records. Orbit paths illustrate the geometry.

An archive of the planet

Air trapped in Antarctic ice records atmospheres some 800,000 years old4. The ice sheet also holds enough water to raise global sea level by about 58 metres8. The Southern Ocean, the cold ocean that circles the continent, takes up roughly 40% of the carbon dioxide and three quarters of the excess heat that the oceans absorb from human activity9. The research stations that study this ice therefore work on questions about climates and coastlines far beyond Antarctica.

A place people already use

Tourism reached about 122,000 visitors in the 2023–24 season5. At sea, fishing vessels catch krill, small shrimp-like animals at the base of the Antarctic food web. The catch supplies fish-farm feed and omega-3 supplements. In 2025 the fishery reached its agreed 620,000-tonne catch limit and closed early7. Through farmed fish, the Southern Ocean already reaches dinner plates far from the ice.

A vantage point

Most Earth-observation satellites fly polar orbits and pass over both poles on every lap. A ground station near the pole, an antenna that downloads their data, can therefore reach them on nearly every pass. That makes Antarctic ground stations valuable for watching the Earth, and for watching each other10. Beneath the ice lie potential mineral resources, which a mining ban keeps off limits6. Scientific cooperation takes place alongside these strategic interests.

All of these interests meet on a continent that no single country owns. Someone has to decide where tourists may land, how much krill may be caught, and whether anyone may mine. Who decides, and how?

Rules Without a Ruler

In 1959, at the height of the Cold War, twelve countries signed the Antarctic Treaty. Seven of them claimed slices of the continent as national territory, and three of those claims overlapped. Some feared that Antarctica could become a testing ground for nuclear weapons. The treaty froze the claims without settling the disputes. Countries kept their existing claims, and the treaty bars new claims. Nothing a country does while the treaty is in force can strengthen or weaken a claim. It also set the continent aside for peace and science11,38,39. In my reading, it was born of a need for peace, in a world sick of death and destruction on a planetary scale.

Besides freezing the claims, the treaty banned military activity and nuclear explosions. On those terms, countries that disagreed deeply about much else, among them the United States and the Soviet Union, could cooperate11. In effect, it was one of the first arms-control agreements of the Cold War40.

The treaty itself has just fourteen articles, and most of what governs Antarctica today builds on those articles. The treaty and the agreements built on it are together called the Antarctic Treaty System. The countries that joined the treaty are known as the Parties. They added separate binding agreements, called conventions, on seals in 1972 and marine living resources in 1980. The 1980 convention has its own commission, which meets every year in Hobart, Australia, and sets the catch limits for krill and fish. In 1988 they finished negotiating a convention to regulate mining that never took legal effect. Three years later they adopted a Protocol on Environmental Protection, an addition to the treaty that I will call the Environmental Protocol. It bans mining outright and designates Antarctica “a natural reserve, devoted to peace and science”41,6. They had written the rules for mining, and then they ruled mining out altogether.

The year 2048 is often quoted as the date when the mining ban expires. The ban has no expiry date. From 2048, any of the countries that make the decisions can ask for a conference to review the Environmental Protocol. The ban stays in force unless the Parties agree to change the ban. Even then, they can lift it only when binding rules for mining are in place6.

The Parties meet roughly once a year at the Antarctic Treaty Consultative Meeting. Twenty-nine of them, the Consultative Parties, make the decisions. They are the original signatories and the states that joined them after conducting substantial research in Antarctica. They decide by consensus, which means that a proposal passes without a vote if none of them objects. Other participants contribute science, expertise, and proposals, but do not decide. Among them are Non-Consultative Parties, countries that have joined the treaty without that research presence. Observers and invited experts also attend, such as the Scientific Committee on Antarctic Research (SCAR), which coordinates Antarctic science, and the association of tour operators42,43.

The Consultative Parties host the meeting in turn, in the alphabetical order of their English names. The 2025 meeting in Milan lasted ten days. Delegations work in English, French, Russian and Spanish, with interpreters. The Committee for Environmental Protection, which advises the meeting on the environment, usually sits during the first week and presents its advice in the second. Delegates comment on drafts of the final report before the meeting adopts it at the end, so the record of each discussion is a negotiated text too45,44,43,48.

Meeting by meeting, those decisions have built a rulebook on top of the treaty. The meetings produce numbered texts, called instruments, which can be Measures, Resolutions or Decisions. Measures bind the Parties once each government has approved them at home. Resolutions urge action, and Decisions handle the meetings’ own business.

Nations bring their interests in territory, science, fisheries and tourism to these meetings. They have kept the claims frozen and the continent demilitarised while writing rules that affect what happens on the ice. Some of those rules reach across the whole continent. The Environmental Protocol, for example, required every dog to leave by April 19946. Others are local, such as the permit that anyone needs before entering a protected area.

This is the system I study. The participants write papers about the problems they care about and build new rules on old ones. Each rule they adopt changes the terms on which they negotiate the next. When a proposal comes up, they can support it, stay silent or object. Attention, decisions and agreement are the lenses I use to study the meeting records.

Attention Has a Neighbourhood

Before anything can be decided, it has to be noticed. Between 1961 and 2025, 66 participants, countries and organisations, submitted 6,573 Working and Information Papers on problems they wanted discussed. Working Papers ask the meeting to act. Information Papers report on a topic. The treaty’s Secretariat, its permanent office, files each paper under one or more of 45 concerns, or topics. Our draft study uses these papers to trace where attention goes28.

A Network of Concerns

A participant specialises in a concern when that concern makes up a larger share of its papers than it does across everyone’s papers. Two concerns sit close when the same participants tend to specialise in both. The idea comes from economics. There the “product space” maps which exports tend to go together, on the reasoning that a country exporting both has the skills to make both16. Replace countries with participants and products with concerns, and the same logic yields a map of the Antarctic agenda.

Science and procedure sit near the centre of the network, with clusters for environmental management, tourism and safety, and resources. Each dot is a concern, and each link shows how related two concerns are. Consultative Parties, Non-Consultative Parties and organisations tend to specialise in different parts of the network.

Hover or focus a concern for its name. Click or tap it to explore its connections, or use the region filters.

45 concerns linked by shared specialisation, in their original positionsArea Protection and Management Plans General · Environmental managementBiological Prospecting · Procedural/scienceCEP Strategy Discussions · Environmental managementClimate Change · Environmental managementComprehensive Environmental Evaluations · Environmental managementCooperation with Other Organisations · Cooperation/reportingDrilling · Resources/planningEducational issues · Procedural/scienceEmergency report and contingency planning · Access/tourism/safetyEnvironmental Domains Analysis · Domains/reportingEnvironmental Impact Assessment EIA Other EIA Matters · Procedural/scienceEnvironmental Monitoring and Reporting · Procedural/scienceEnvironmental Protection General · Procedural/scienceExchange of Information · Procedural/scienceFauna and Flora_General · Environmental managementHistoric Sites and Monuments · Environmental managementHuman Footprint and wilderness values · Environmental managementInspections · Procedural/scienceInstitutional and legal matters · Environmental managementInternational Polar Year · Procedural/scienceLiability · Procedural/scienceManagement Plans · Environmental managementMarine Acoustics · Acoustics/lakesMarine Protected Areas · Access/tourism/safetyMarine living resources · Resources/planningMineral resources · Resources/planningMultiyear strategic workplan · Resources/planningNonnative Species and Quarantine · Environmental managementOpening statements · Procedural/scienceOperation of the Antarctic Treaty system General · Procedural/scienceOperation of the Antarctic Treaty system Reports · Cooperation/reportingOperation of the Antarctic Treaty system The Secretariat · Procedural/scienceOperation of the CEP · Environmental managementOperational issues · Procedural/sciencePrevention of marine pollution · Access/tourism/safetyRepair and remediation of environmental damage · Environmental managementSafety and Operations in Antarctica · Procedural/scienceScience issues · Procedural/scienceSearch and Rescue · Access/tourism/safetySite Guidelines for Visitors · Access/tourism/safetySpecially Protected Species · Environmental managementState of the Antarctic Environment Report SAER · Domains/reportingSub glacial Lakes · Acoustics/lakesTourism and NG_Activities · Access/tourism/safetyWaste management and disposal · Procedural/science

When participants take up a new concern, it tends to lie near the ones they already work on. The rate at which participants start to specialise in a concern falls from 9.4% for the nearest concerns to 2.7% for the farthest1. This pattern is called path dependence. What a participant has worked on limits what it is likely to take up next. I think new concerns are most likely to reach the agenda through participants who already work on related issues28.

New specialisation falls from 9.4% to 2.7% across distance decilesAll ten observed distance groups with Wilson 95 percent intervals. Distance is one minus the strongest co-specialisation with the earlier portfolio.New specialisation (%)048129.43% at distance 0.4119.09% at distance 0.5016.77% at distance 0.5476.62% at distance 0.5865.38% at distance 0.6205.11% at distance 0.6555.22% at distance 0.7013.73% at distance 0.7422.68% at distance 0.8032.72% at distance 0.9019.4%2.7%0.40.60.8Near earlier workFar from earlier work →
Distance from a participant’s previous portfolio (1 − maximum co-specialisation). Points show the original ten groups, and bars show Wilson 95% intervals.

What participants write about helps explain the subjects of agreements, beyond what earlier agreements alone reveal. The signal is clearest for non-binding Resolutions, while Decisions show no clear improvement. The comparisons use papers from the meeting we evaluate and from the previous meeting. They tell us how much the papers explain about meetings we have already observed. The comparison leaves open whether the papers can predict what later meetings adopt28.

A Reference Is Not a Change

In our draft study, we used AI language models to trace links among 793 numbered instruments and see how decisions build on earlier ones. We took the instruments from the final reports of 47 meetings held in 1961–2025. The final report is the official record of a meeting29.

The Decision Lineages of Antarctic Governance

We use the treaty records to trace when the Parties adopt an instrument, what led to its adoption and how later instruments refer to or change the text. Chains of instruments that act on earlier ones form lineages.

With references included, most instruments belong to one connected network. When we keep only links where an instrument amends, replaces or retires an earlier one, which I call actions, we find 46 lineages. They cover budget matters and conservation efforts. Among the conservation efforts are protected areas, sites that need a permit to enter because of their scientific or environmental value, and management plans, the rules for each site.

A reader who knows the treaty well might recognise those links across documents. Putting them in a network lets us study how the rules grew across meetings.

Switch the links. Hover or focus an instrument for its name. Click or tap to explore its connections. Every instrument stays in the same position.

Measure 1 (2024)Measure 2 (2024)Measure 3 (2024)Measure 4 (2024)Measure 5 (2024)Measure 6 (2024)Measure 7 (2024)Measure 8 (2024)Measure 9 (2024)Measure 10 (2024)Measure 11 (2024)Measure 12 (2024)Measure 13 (2024)Measure 14 (2024)Measure 17 (2024)Measure 18 (2024)Decision 1 (2024)Decision 2 (2024)Decision 4 (2024)Resolution 1 (2024)Resolution 2 (2024)Measure 1 (2025)Measure 2 (2025)Measure 3 (2025)Decision 1 (2025)Decision 2 (2025)Decision 6 (2025)Decision 7 (2025)Resolution 1 (2025)Resolution 2 (2025)Resolution 3 (2025)Resolution 4 (2025)Recommendation 3-3Recommendation 3-4Recommendation 3-5Recommendation 3-6Recommendation 3-8Recommendation 3-9Recommendation 3-10Recommendation 2-2Recommendation 2-3Recommendation 2-5Recommendation 2-8Recommendation 2-10Recommendation 4-1Recommendation 4-2Recommendation 4-3Recommendation 4-4Recommendation 4-5Recommendation 4-6Recommendation 4-7Recommendation 4-8Recommendation 4-9Recommendation 4-10Recommendation 4-11Recommendation 4-12Recommendation 4-13Recommendation 4-14Recommendation 4-16Recommendation 4-17Recommendation 4-19Recommendation 4-20Recommendation 4-25Recommendation 4-26Recommendation 4-28Recommendation 9-3Recommendation 9-4Recommendation 9-5Recommendation 9-6Recommendation 1-7Recommendation 1-8Recommendation 1-9Recommendation 1-11Recommendation 1-15Recommendation 8-1Recommendation 8-2Recommendation 8-3Recommendation 8-4Recommendation 8-5Recommendation 8-6Recommendation 8-7Recommendation 8-9Recommendation 8-11Recommendation 8-13Recommendation 7-1Recommendation 7-2Recommendation 7-3Recommendation 7-5Recommendation 7-7Recommendation 7-9Recommendation 6-1Recommendation 6-2Recommendation 6-3Recommendation 6-4Recommendation 6-8Recommendation 6-9Recommendation 6-14Recommendation 6-15Recommendation 5-1Recommendation 5-2Recommendation 5-4Recommendation 5-5Recommendation 5-8Recommendation 5-9Recommendation 13-3Recommendation 13-4Recommendation 13-5Recommendation 13-7Recommendation 13-8Recommendation 13-9Recommendation 13-10Recommendation 13-11Recommendation 13-12Recommendation 13-13Recommendation 13-14Recommendation 13-16Recommendation 12-1Recommendation 12-2Recommendation 12-3Recommendation 12-4Recommendation 12-5Recommendation 12-6Recommendation 12-7Recommendation 14-1Recommendation 14-2Recommendation 14-4Recommendation 14-5Recommendation 14-6Recommendation 14-7Recommendation 14-8Recommendation 14-9Recommendation 14-10Measure 1 (1995)Measure 2 (1995)Measure 3 (1995)Measure 4 (1995)Measure 5 (1995)Resolution 1 (1995)Resolution 4 (1995)Resolution 5 (1995)Resolution 6 (1995)Resolution 7 (1995)Recommendation 11-2Measure 1 (2021)Measure 2 (2021)Measure 3 (2021)Measure 4 (2021)Measure 5 (2021)Measure 6 (2021)Measure 7 (2021)Measure 8 (2021)Measure 9 (2021)Measure 10 (2021)Measure 11 (2021)Measure 12 (2021)Measure 13 (2021)Measure 14 (2021)Measure 15 (2021)Measure 16 (2021)Measure 17 (2021)Measure 18 (2021)Measure 23 (2021)Decision 2 (2021)Decision 3 (2021)Decision 5 (2021)Decision 7 (2021)Resolution 2 (2021)Resolution 3 (2021)Resolution 4 (2021)Resolution 6 (2021)Resolution 10 (2021)Measure 1 (2019)Measure 2 (2019)Measure 3 (2019)Measure 4 (2019)Measure 5 (2019)Measure 6 (2019)Measure 7 (2019)Measure 8 (2019)Measure 9 (2019)Measure 10 (2019)Measure 11 (2019)Measure 12 (2019)Decision 1 (2019)Decision 2 (2019)Decision 4 (2019)Decision 5 (2019)Decision 7 (2019)Resolution 1 (2019)Resolution 2 (2019)Resolution 3 (2019)Measure 1 (2022)Measure 2 (2022)Measure 3 (2022)Measure 4 (2022)Measure 5 (2022)Measure 6 (2022)Measure 7 (2022)Measure 8 (2022)Measure 9 (2022)Measure 10 (2022)Measure 11 (2022)Measure 12 (2022)Measure 13 (2022)Measure 14 (2022)Measure 15 (2022)Measure 16 (2022)Measure 17 (2022)Measure 18 (2022)Decision 2 (2022)Decision 3 (2022)Decision 5 (2022)Resolution 1 (2022)Resolution 2 (2022)Resolution 3 (2022)Resolution 6 (2022)Measure 1 (2018)Measure 2 (2018)Measure 3 (2018)Measure 4 (2018)Measure 5 (2018)Measure 6 (2018)Decision 3 (2018)Resolution 1 (2018)Resolution 2 (2018)Resolution 3 (2018)Measure 1 (2023)Measure 2 (2023)Measure 3 (2023)Measure 4 (2023)Measure 5 (2023)Measure 6 (2023)Measure 7 (2023)Measure 8 (2023)Measure 9 (2023)Measure 10 (2023)Measure 11 (2023)Measure 12 (2023)Measure 13 (2023)Measure 14 (2023)Measure 15 (2023)Measure 16 (2023)Measure 17 (2023)Decision 1 (2023)Decision 4 (2023)Decision 5 (2023)Measure 1 (2017)Measure 2 (2017)Measure 3 (2017)Measure 4 (2017)Measure 5 (2017)Measure 6 (2017)Measure 7 (2017)Measure 8 (2017)Decision 2 (2017)Decision 3 (2017)Decision 7 (2017)Resolution 3 (2017)Resolution 6 (2017)Recommendation 17-1Recommendation 17-2Recommendation 17-3Recommendation 17-4Recommendation 16-2Recommendation 16-3Recommendation 16-4Recommendation 16-5Recommendation 16-6Recommendation 16-7Recommendation 16-8Recommendation 16-9Recommendation 16-10Recommendation 16-11Recommendation 16-13Recommendation 15-1Recommendation 15-2Recommendation 15-3Recommendation 15-4Recommendation 15-6Recommendation 15-7Recommendation 15-8Recommendation 15-9Recommendation 15-10Recommendation 15-11Recommendation 15-12Recommendation 15-13Recommendation 15-16Recommendation 15-18Recommendation 15-19Recommendation 15-20Recommendation 15-22Measure 1 (1999)Decision 2 (1999)Resolution 1 (1999)Resolution 4 (1999)Measure 1 (1998)Measure 2 (1998)Decision 2 (1998)Decision 3 (1998)Decision 4 (1998)Resolution 1 (1998)Resolution 2 (1998)Resolution 6 (1998)Measure 1 (2001)Measure 2 (2001)Measure 3 (2001)Decision 2 (2001)Decision 3 (2001)Measure 1 (2006)Measure 2 (2006)Measure 3 (2006)Decision 2 (2006)Resolution 2 (2006)Measure 1 (1997)Measure 2 (1997)Measure 3 (1997)Measure 4 (1997)Measure 5 (1997)Decision 2 (1997)Resolution 1 (1997)Measure 2 (2005)Measure 3 (2005)Measure 5 (2005)Decision 1 (2005)Decision 3 (2005)Decision 4 (2005)Decision 6 (2005)Decision 7 (2005)Decision 8 (2005)Decision 9 (2005)Resolution 3 (2005)Resolution 4 (2005)Resolution 5 (2005)Resolution 6 (2005)Measure 1 (2004)Measure 2 (2004)Measure 3 (2004)Measure 4 (2004)Decision 1 (2004)Decision 2 (2004)Resolution 4 (2004)Measure 2 (2003)Measure 3 (2003)Decision 2 (2003)Decision 3 (2003)Decision 4 (2003)Resolution 3 (2003)Measure 1 (2002)Decision 3 (2002)Measure 1 (2010)Measure 2 (2010)Measure 3 (2010)Measure 4 (2010)Measure 5 (2010)Measure 6 (2010)Measure 7 (2010)Measure 8 (2010)Measure 9 (2010)Measure 10 (2010)Measure 11 (2010)Measure 12 (2010)Measure 13 (2010)Measure 14 (2010)Measure 15 (2010)Decision 3 (2010)Decision 4 (2010)Resolution 1 (2010)Resolution 2 (2010)Resolution 3 (2010)Measure 1 (2009)Measure 2 (2009)Measure 3 (2009)Measure 4 (2009)Measure 5 (2009)Measure 6 (2009)Measure 7 (2009)Measure 8 (2009)Measure 9 (2009)Measure 10 (2009)Measure 11 (2009)Measure 12 (2009)Measure 13 (2009)Measure 14 (2009)Decision 2 (2009)Decision 3 (2009)Decision 4 (2009)Decision 6 (2009)Resolution 4 (2009)Resolution 8 (2009)Measure 1 (2011)Measure 2 (2011)Measure 3 (2011)Measure 4 (2011)Measure 5 (2011)Measure 6 (2011)Measure 7 (2011)Measure 8 (2011)Measure 9 (2011)Measure 10 (2011)Measure 11 (2011)Measure 12 (2011)Decision 1 (2011)Decision 2 (2011)Resolution 1 (2011)Resolution 2 (2011)Resolution 3 (2011)Resolution 4 (2011)Resolution 5 (2011)Measure 1 (2016)Measure 2 (2016)Measure 3 (2016)Measure 4 (2016)Measure 5 (2016)Measure 6 (2016)Measure 7 (2016)Measure 8 (2016)Decision 2 (2016)Decision 5 (2016)Decision 6 (2016)Resolution 1 (2016)Resolution 2 (2016)Resolution 5 (2016)Measure 1 (2008)Measure 2 (2008)Measure 3 (2008)Measure 4 (2008)Measure 5 (2008)Measure 6 (2008)Measure 7 (2008)Measure 8 (2008)Measure 9 (2008)Measure 10 (2008)Measure 11 (2008)Measure 12 (2008)Measure 13 (2008)Measure 14 (2008)Decision 1 (2008)Decision 3 (2008)Resolution 1 (2008)Resolution 2 (2008)Resolution 5 (2008)Measure 1 (2015)Measure 2 (2015)Measure 3 (2015)Measure 4 (2015)Measure 5 (2015)Measure 6 (2015)Measure 7 (2015)Measure 8 (2015)Measure 9 (2015)Measure 10 (2015)Measure 11 (2015)Measure 12 (2015)Measure 13 (2015)Measure 14 (2015)Measure 15 (2015)Measure 16 (2015)Measure 17 (2015)Measure 18 (2015)Decision 1 (2015)Decision 2 (2015)Decision 4 (2015)Decision 5 (2015)Decision 6 (2015)Measure 1 (2014)Measure 2 (2014)Measure 3 (2014)Measure 4 (2014)Measure 5 (2014)Measure 6 (2014)Measure 7 (2014)Measure 8 (2014)Measure 9 (2014)Measure 10 (2014)Measure 11 (2014)Measure 13 (2014)Measure 14 (2014)Measure 15 (2014)Measure 16 (2014)Decision 1 (2014)Decision 3 (2014)Resolution 4 (2014)Resolution 7 (2014)Measure 1 (2013)Measure 2 (2013)Measure 3 (2013)Measure 4 (2013)Measure 5 (2013)Measure 6 (2013)Measure 7 (2013)Measure 8 (2013)Measure 9 (2013)Measure 10 (2013)Measure 11 (2013)Measure 12 (2013)Measure 13 (2013)Measure 14 (2013)Measure 15 (2013)Measure 16 (2013)Measure 17 (2013)Measure 18 (2013)Measure 19 (2013)Measure 20 (2013)Measure 21 (2013)Decision 5 (2013)Decision 6 (2013)Decision 7 (2013)Resolution 1 (2013)Resolution 2 (2013)Resolution 3 (2013)Measure 1 (2012)Measure 2 (2012)Measure 3 (2012)Measure 5 (2012)Measure 6 (2012)Measure 7 (2012)Measure 8 (2012)Measure 9 (2012)Measure 10 (2012)Measure 11 (2012)Decision 1 (2012)Decision 2 (2012)Decision 3 (2012)Resolution 1 (2012)Resolution 4 (2012)Resolution 6 (2012)Measure 1 (2007)Measure 2 (2007)Measure 3 (2007)Decision 1 (2007)Decision 3 (2007)Resolution 1 (2007)Measure 1 (1996)Measure 2 (1996)Resolution 1 (1996)Resolution 3 (1996)Resolution 5 (1996)Recommendation 10-3Recommendation 10-4Recommendation 10-5Recommendation 10-6Recommendation 10-7Recommendation 10-8Recommendation 10-9Measure 1 (2000)Measure 2 (2000)Resolution 1 (2000)Recommendation 1-5Recommendation 4-21Recommendation 6-5Recommendation 8-12Recommendation 16-1Resolution 2 (1996)Decision 1 (1999)Resolution 6 (1999)Resolution 2 (2000)Decision 2 (2002)Resolution 1 (2003)Resolution 4 (2003)Measure 4 (2006)Resolution 2 (2007)Measure 16 (2009)Resolution 1 (2009)Resolution 5 (2009)Resolution 9 (2009)Resolution 4 (2010)Resolution 3 (2012)Resolution 5 (2013)Resolution 3 (2015)Resolution 6 (2016)Resolution 4 (2017)Resolution 7 (2021)Recommendation 1-14Measure 15 (2009)Resolution 4 (2023)Resolution 2 (2012)Recommendation 1-2Resolution 6 (2019)Resolution 5 (2014)Measure 12 (2014)Resolution 2 (2014)Resolution 1 (2014)Recommendation 18-1Resolution 5 (2012)Resolution 2 (2015)Decision 4 (2017)Resolution 1 (2006)Resolution 5 (2017)Resolution 1 (2015)Recommendation 7-8Recommendation 13-1Decision 2 (1995)Decision 1 (1997)Resolution 5 (1998)Resolution 4 (2015)Resolution 6 (2015)Decision 1 (2017)Resolution 8 (2021)Resolution 4 (2022)Decision 4 (2022)Resolution 2 (2023)Recommendation 1-16Resolution 9 (2012)Resolution 5 (1999)Resolution 4 (2007)Resolution 5 (2019)Resolution 9 (2021)Resolution 5 (2025)Resolution 4 (2019)Measure 21 (2021)Measure 16 (2024)Measure 19 (2021)Decision 1 (2002)Resolution 11 (2012)Resolution 4 (2008)Recommendation 16-12Resolution 2 (2005)Recommendation 4-24Decision 5 (2003)Decision 1 (2009)Decision 7 (2009)Measure 4 (2005)Measure 22 (2021)Recommendation 4-15Recommendation 12-8Decision 1 (1995)Resolution 1 (2001)Recommendation 2-7Decision 1 (2013)Decision 1 (1998)Decision 2 (2005)Recommendation 6-10Resolution 2 (2002)Measure 4 (2012)Measure 1 (2005)Recommendation 15-5Recommendation 13-2Recommendation 14-3Recommendation 15-17Recommendation 15-14Recommendation 15-15Recommendation 15-21Recommendation 13-6Recommendation 11-1Recommendation 7-6Resolution 1 (2005)Recommendation 8-14Recommendation 10-1Recommendation 1-4Recommendation 4-22Recommendation 9-1Resolution 6 (2011)Resolution 4 (2016)Recommendation 9-2Recommendation 10-2Recommendation 3-7Recommendation 3-11Recommendation 4-18Recommendation 5-6Resolution 2 (1999)Resolution 1 (2002)Resolution 1 (2023)Resolution 2 (2004)Resolution 4 (2018)Resolution 5 (2007)Resolution 7 (2009)Resolution 6 (2014)Resolution 5 (2022)Decision 1 (2022)Decision 3 (2024)Decision 3 (2025)Decision 4 (2025)Resolution 2 (1997)Decision 5 (2005)Decision 2 (2010)Decision 3 (2011)Decision 3 (2013)Decision 2 (2018)Decision 2 (2023)Decision 3 (2023)Recommendation 11-3Resolution 5 (2001)Resolution 8 (1995)Resolution 4 (2001)Resolution 7 (2019)Decision 1 (2010)Measure 19 (2015)Resolution 4 (1996)Resolution 3 (2009)Decision 1 (2021)Decision 1 (2006)Decision 4 (2007)Decision 2 (2008)Recommendation 8-10Decision 6 (2019)Decision 6 (2021)Resolution 3 (2006)Measure 9 (2016)Recommendation 6-11Recommendation 7-4Recommendation 1-10Resolution 6 (2001)Resolution 5 (2004)Decision 4 (2013)Decision 2 (2014)Decision 3 (2015)Decision 3 (2016)Decision 5 (2017)Decision 1 (2018)Resolution 3 (2016)Resolution 3 (2001)Resolution 5 (2018)Resolution 1 (2021)Measure 18 (2023)Measure 15 (2024)Decision 10 (2005)Decision 4 (2012)Decision 6 (2023)Decision 5 (2024)Resolution 1 (2017)Decision 4 (2004)Resolution 7 (2010)Resolution 3 (2014)Decision 1 (2001)Decision 1 (2003)Decision 2 (2007)Decision 4 (2008)Decision 5 (2009)Resolution 3 (1998)Decision 3 (2004)Resolution 9 (1995)Resolution 3 (2008)Measure 1 (2003)Decision 2 (2013)Recommendation 2-9Decision 4 (2016)Resolution 5 (2015)Measure 20 (2021)Recommendation 4-27Recommendation 1-6Recommendation 3-1Recommendation 4-23Decision 4 (2021)Decision 5 (2025)Decision 3 (2019)Decision 6 (2017)Resolution 3 (1995)Resolution 3 (1997)Decision 5 (2008)Recommendation 1-1Recommendation 1-3Recommendation 1-13Recommendation 2-1Recommendation 2-4Recommendation 2-6Recommendation 3-2Recommendation 5-3Recommendation 6-6Recommendation 6-7Recommendation 6-12Recommendation 6-13Resolution 3 (2004)Resolution 1 (2004)Resolution 6 (2008)Resolution 6 (2010)Resolution 7 (2012)Resolution 8 (2012)Resolution 4 (2013)Recommendation 1-12Recommendation 5-7Recommendation 8-8Recommendation 13-15Resolution 2 (1995)Resolution 4 (1998)Resolution 3 (1999)Decision 1 (2000)Resolution 2 (2001)Resolution 3 (2002)Resolution 2 (2003)Resolution 7 (2005)Resolution 4 (2006)Resolution 3 (2007)Decision 8 (2009)Resolution 2 (2009)Resolution 6 (2009)Decision 5 (2010)Resolution 5 (2010)Resolution 10 (2012)Resolution 6 (2013)Decision 1 (2016)Resolution 2 (2017)Resolution 5 (2021)Resolution 3 (2023)

With references included, 724 of the 793 instruments are connected to each other, directly or through other instruments. With only actions included, 244 instruments have no link at all. Many links that look like actions are not. The reports contain 943 records of an instrument adopting or designating something that an earlier instrument covered. In 916 of them the wording is a recital, one of the opening “recalling…” lines of a text. Those lines refer to earlier instruments without acting on them29.

Instruments usually act on others of the same type. Measures rarely act on Decisions or Resolutions, and 239 of the 243 actions on binding Measures come from other Measures. Recommendations are the exception. They were the only instrument type until Measures, Resolutions and Decisions replaced Recommendations in 1995. The difference matters because Measures bind once the Parties approve them at home. That gives the rules legal force in a shared region without a government. Resolutions can guide action without binding the Parties. In the records we analysed since 1995, no Resolution became a binding Measure29.

Who changes a binding Measure?239 / 243
Actions by other Measures4 by other types

Operative pairs in the instrument dataset

What are binding Measures about?257 / 263
Protected areas, plans, and historic sites6 on other subjects

Measures adopted in 1999–2025

Both counts show that the binding core mostly maintains existing rules and the places they govern. Operative relationships still need an independent human audit.

The binding core of the rulebook is concentrated on places. Of the 263 Measures adopted between 1999 and 2025, 257 concern protected areas, their management plans or historic sites30. New Measures mostly maintain that core. Most site plans adopted in 1995–2012, for example, received their first revision after five or six years, close to the five-year review that the Environmental Protocol requires. Only 8 of 84 went untouched for twelve years29. The timing suggests that the Parties generally follow the review schedule.

Much agreement, however, never becomes a numbered rule at all. About a quarter of the Working Papers (425 of 1,595) were agreed or endorsed without producing any instrument29. To understand what the meetings decide, we therefore have to look at agreement itself.

The Same “Yes” Can Hide Different Positions

As the sociologist Philippe Urfalino has argued, consensus operates through non-opposition, so a proposal can pass when nobody objects19. The record can look the same whether a Party endorses a proposal or stands aside to let it pass. A record of agreement therefore tells us little about how much support the proposal had.

Different responses can leave the same record of agreement

One proposalEach participant chooses a response.

Participant A

Supports the proposal

Supports
Participant B

Blocking brings no immediate gain

Stands aside
Participant C

Supports the proposal

Supports
AgreementThe proposal passes with one participant standing aside.
Choose an example. Different positions can produce the same agreement record. These three participants illustrate the decision rule and possible responses, without estimating particular countries' motives.

A Party may stand aside because it favours a proposal too weakly to endorse publicly or because blocking would bring no immediate gain. Future cooperation can also be a reason to stand aside. A proposer can reduce its demands before or after an objection. A revision can win active support or make a proposal tolerable enough for a Party to let it pass. When a proposal passes while at least one participant stands aside, we call that apparent consensus. That participant may still favour the proposal31.

Records of agreement and numbered instruments do not always occur together. In our draft study of agreement records, we compared 180 agenda items from 1999–2025 and labelled what the record says about each. In 73.3% the record says that the meeting adopted something. A numbered document appears in 51.7%, and a Decision or Measure in 33.3%. These labels describe different things about the same items. Nine agenda items, for instance, combine a record of no consensus with a numbered document. An agenda item can cover several proposals, so the label and the document may concern different matters30.

Does Agreement Mean Agreeing?

In our draft study of consensus, we built a bargaining game, a simple computer model with two (or more) simulated participants. They meet round after round and take turns making a proposal, which can ask for a lot or a little. The other participant chooses whether to support the proposal, stand aside or use a veto. In ordinary play, vetoing a demanding proposal leads to a revision round, where only a modest proposal is on offer. Other vetoes lead to a dispute. In a dispute, called a punishment phase, passing a proposal gives the pair a chance to recover. After each round the game continues with a fixed chance. That chance tells the participants how much future they have together.

The three responses show alternative choices by the same participant. In ordinary talks, vetoing a demanding proposal opens a revision round. Other vetoes lead to a dispute, where passage gives a chance to recover.

We let 1,000 pairs learn the game by trial and error for two million rounds each, and then studied their final 4,000 rounds. When the pair almost always meets again, with a 23/24 chance of another round, 95.1% of rounds end in agreement. Most of that agreement is apparent consensus. Someone stands aside in 83.5% of rounds, and the proposal receives full support in 11.5%31.

We also ask how secure that agreement is. We turn each participant’s most likely choices into a rule and ask whether either could gain by choosing differently while the other follows its rule. In the situations that ordinary play reaches, 889 of the 1,000 pairs pass, because neither participant has a reason to change a choice. The full check also covers choices after a disagreement and responses to proposals the pair usually avoids. Only three pairs pass that full check. Punishment accounts for just 0.1% of the final training rounds, so the pairs have had almost no practice responding to a dispute31.

The chance of meeting again matters. When we lower it to 5/6, participants stand aside less often and far more pairs pass the full check2. We also restarted trained pairs in the middle of a punishment phase, to see how a dispute affects their behaviour. Many participants changed how they respond to a dispute, while the rate of apparent consensus stayed the same331.

I think the model shows how participants can keep agreeing while having reasons to change their choices after a dispute. Restarting them after a dispute changes how they respond, even when the agreement rate barely moves. The incentive checks and restart experiment leave open how likely a real agreement is to collapse and what motivates a country to stand aside.

Hard Rules for Places, Soft Rules for Problems

Participants tend to take up concerns close to the ones they already work on. Binding rules mostly update other binding rules, and countries can let a proposal pass by standing aside even when they disagree. That helps explain how the Parties can keep rules for particular places working while they discuss a much wider set of problems. What participants write about shows up most clearly in non-binding Resolutions, and a quarter of Working Papers win agreement without producing an instrument. The practices that keep the treaty working can also leave difficult problems unresolved.

In our draft study of consensus, we compared 366 Working Papers that proposed revisions to existing arrangements or new arrangements. The meeting that received the paper adopted an instrument for 220 of 276 revisions (79.7%), compared with 34 of 90 new proposals (37.8%). After adjusting for how a proposal moved through the agenda, the period and whether it concerned a particular site, the difference is 28 percentage points (95% interval 17 to 41). Revisions pass more readily at first. Whether new proposals catch up at later meetings is still open4. We identified the patterns after assembling the records, so a proper test will need new observations31.

Annex VI to the Environmental Protocol would fill a gap in the treaty’s rules on liability, the question of who pays for environmental damage. It would require the national programmes and companies that run expeditions to prevent and respond to environmental emergencies. It would also make them pay when they fail to respond. Negotiating it took years, with recorded failures to reach consensus in 1998 and 2003, before it was adopted as Measure 1 (2005). Like every Measure, it could not take effect until all 28 countries that were Consultative Parties in 2005 had approved it at home. By 2024, only 19 had done so, and the annex had not taken effect36.

Ferrada and Caldera, who studied the negotiation, point to disagreements over the annex’s scope, potential costs to states, and Antarctica’s low place among domestic priorities. The annex shows how a meeting can reach agreement without securing the approvals needed to make it effective36. In 2018, another scholar, Hemmings, found no one campaigning for the annex and wondered whether it would take a catastrophe for anyone to start35. A third, Bastmeijer, calls the delay a “success syndrome”. In his reading, the meetings struggle to meet the expectations that the success of the Environmental Protocol created37. If he is right, the annex may only take effect after an emergency of the kind it was written for.

Viewed through its work on science, conservation and logistics, the Antarctic Treaty System looks orderly, even dull. It is also a place where countries compete for influence, often without acting on that competition directly. Cases the record does not explain bring that tension into view.

The emperor penguin is such a case. A proposal to protect the species stalled before agreement. Climate change is eroding the sea ice on which the species breeds. In 2022 the United Kingdom, building on advice from the scientific committee SCAR, proposed designating it a Specially Protected Species, a status reserved for species that need extra protection. No Party harvests the penguin, and the attention network shows environmental management among the busiest parts of the agenda. The proposal nevertheless remained unagreed through 2024. The 2022 record names China, which “reiterated its position that the emperor penguin was not currently eligible for such a designation”. The 2024 record refers to “some Parties” and says “the Meeting could not reach consensus”32,33.

In 2025, China co-proposed the revised management plan for the Amanda Bay emperor penguin colony, and the plan passed. China continued to oppose protection for the species as a whole. The plan updates protection first adopted in 2008. It is a separate proposal, so its passage does not show that the species proposal was weakened until China accepted the text48. In 2026, China again considered species designation premature, and Russia supported China’s statement. The meeting agreed to continue considering appropriate protection measures at its next meeting49.

China gave scientific reasons for opposing species designation. The record leaves open whether strategic considerations also played a part. Granting enhanced protection on the basis of projected rather than current threats could set a precedent for restrictions elsewhere, for instance on the fisheries that several Parties pursue in the Southern Ocean. Conservation rules can also be read as a way for some Parties to constrain how others act in the region, an echo of the sovereignty question the treaty froze but never resolved.

These lenses cannot settle the case, but they suggest where to look. With the attention lens, we can ask how conservation relates to the concerns China already works on. With the decision lens, we can distinguish revisions to the Amanda Bay plan from the proposed protection of the species. The history shows that a Party can support narrower site protection while opposing species designation. Passage still leaves each Party’s reasons for allowing the plan unresolved.

Networks can point to records worth reading. To understand why a Party objects, we still have to read what it said in the context of diplomacy and law.

What the Record Can Tell Us

It takes years of records to see how participants take up new concerns, build on earlier decisions and let proposals pass. These patterns help describe how the institution persists and changes. They do not, on their own, tell us whether its decisions are implemented or achieve their intended effects. Persistence, adaptability and effectiveness are different questions, even if they are often discussed as though they were the same.

That distinction matters for judging the treaty itself. Some studies question the effectiveness of international treaties, since formally joining them often changes little about what states actually do22,23. The Parties adopted Annex VI, but nearly two decades passed without it taking effect. Other records show a system that repeatedly returns to its rules, revises them and accommodates divergent interests without open conflict at the meetings. Those findings complicate a simple account of failure, but they do not establish that the rules work as intended on the ground. An institution can remain active without being effective.

The record itself is also a compressed account of what happened25. A citation tells us that one instrument mentions another, but the later instrument may leave the earlier rule unchanged. An adoption records that a proposal passed, while leaving open whether everyone supported the proposal. Recorded disagreement tells us when disagreement was written down, but leaves open how much went unspoken.

Computation lets us examine sixty years of documents, but extracting relationships from text introduces uncertainty. When other language models checked our extracted links, they accepted between roughly two in five and five in six, depending on the link type. These cross-model acceptance rates do not establish accuracy independently, so human validation remains necessary.

Explaining why a Party objects or stands aside will require connecting the records to national interests, proposals and the reasoning of the people in the room. Annex VI took years to negotiate and still awaits the domestic approvals needed to take effect. The emperor penguin species proposal remained unagreed in 2026, even as the separate Amanda Bay plan was revised. The records give reasons, including costs, scope and scientific objections, but they cannot settle every strategic explanation. Clearer records would help us study disagreement and hold decision-makers accountable.

I focused on the people at the table, but Antarctica’s ecosystems and climate are complex systems too. Some of the most interesting questions may concern how these systems interact with the people who govern the continent.

The planetary problems I started with call for collective action, but national governments also have responsibilities to their own populations. As a Dutch person, I think about that tension in a country where protecting people from rising seas may one day mean difficult decisions about where they can live and who bears the costs. Cooperation across borders cannot make those local responsibilities disappear.

The ecologist Garrett Hardin warned in 1968 that people sharing a resource could overuse it when each pursued their own interests21. The usual alternatives of central control and private ownership are difficult to apply to Antarctica. No state stands above the Parties, and the treaty froze the claims that would divide the continent. The political scientist Elinor Ostrom showed that communities of fishers and farmers can also govern shared resources through rules they devise and monitor themselves20. To me, the Antarctic Treaty System resembles that third way, tried among states.

The Parties left their deepest dispute over sovereignty unsettled, which allowed cooperation on other matters. They began with the shared purposes of peace and science. Their binding rules now concentrate heavily on particular places, where agreement is easier than on more general problems. Existing arrangements are easier to revise than entirely new ones are to adopt. Liability rules and protection for a species threatened by climate change have waited for years. I am cautious about drawing lessons for other shared problems from this one case.

Perhaps several centres of authority can help people retain control over their own affairs while negotiating what they must do together. For an institution that endures, we still need to ask whether it can adapt and whether its decisions change what happens beyond the meeting room. Antarctica gives us a place to study how those demands fit together.

Antarctica will not stop the planet from warming or end the wars on the news. But it is one of the few places where sovereign nations have governed a shared space together for more than sixty years. Among them are the same great powers that sit at the table in every other forum. Understanding how they manage it, and what it costs them, seems to me a good place to start.

And if anyone would like to collaborate (or fund a few years of this research), you can find me through the contact page above!

References

References

  1. 1.

    Uppsala University (2026). UCDP: record number of conflicts between states. Link

  2. 2.

    United Nations Environment Programme (2025). Emissions Gap Report 2025: Off Target. Link

  3. 3.

    Planetary Boundaries Science Lab (2025). Planetary Health Check 2025: A Scientific Assessment of the State of the Planet. Link

  4. 4.

    Lüthi, Dieter, Le Floch, Martine, Bereiter, Bernhard, Blunier, Thomas, Barnola, Jean-Marc, Siegenthaler, Urs, et al. (2008). High-resolution carbon dioxide concentration record 650,000--800,000 years before present. Nature, 453, 379–382. Link

  5. 5.

    International Association of Antarctica Tour Operators (2024). IAATO Overview of Antarctic Vessel Tourism: The 2023-24 Season, and Preliminary Estimates for 2024-25. Link

  6. 6.

    Antarctic Treaty Consultative Parties (1991). Protocol on Environmental Protection to the Antarctic Treaty. Article 7: prohibition of mineral resource activities. Signed Madrid, 4 October 1991; in force 1998

  7. 7.

    Associated Press (2025). Antarctica's krill fishery shut down after record catch triggers early closure. Link

  8. 8.

    Fretwell, P., Pritchard, H. D., Vaughan, D. G., others (2013). Bedmap2: improved ice bed, surface and thickness datasets for Antarctica. The Cryosphere, 7, 375–393. Link

  9. 9.

    Frölicher, Thomas L., Sarmiento, Jorge L., Paynter, David J., Dunne, John P., Krasting, John P., Winton, Michael (2015). Dominance of the Southern Ocean in anthropogenic carbon and heat uptake in CMIP5 models. Journal of Climate, 28(2), 862–886. Link

  10. 10.

    Norwegian Ministry of Trade, Industry (2013). Between Heaven and Earth: Norwegian Space Policy for Business and Public Benefit. Chapter 7: Norway's space-related ground infrastructure. Link

  11. 11.

    Antarctic Treaty Consultative Parties (1959). The Antarctic Treaty. Signed Washington, 1 December 1959; in force 23 June 1961. Link

  12. 12.

    Ladyman, James, Lambert, James, Wiesner, Karoline (2013). What is a complex system?. European Journal for Philosophy of Science, 3(1), 33–67. Link

  13. 13.

    van Elteren, Casper, Quax, Rick, Sloot, Peter (2022). Dynamic importance of network nodes is poorly predicted by static structural features. Physica A: Statistical Mechanics and its Applications, 126889. Link

  14. 14.

    van Elteren, Casper, Vasconcelos, Vítor V., Lees, Mike (2024). Criminal organizations exhibit hysteresis, resilience, and robustness by balancing security and efficiency. Scientific Reports. Link

  15. 15.

    van Elteren, Casper, Vasconcelos, Vítor V., Lees, Mike (2025). The paradox of intervention: Resilience in adaptive multirole coordination networks. Proceedings of the National Academy of Sciences, 122(51), e2509856122. Link

  16. 16.

    Hidalgo, C. A., Klinger, B., Barabási, A.-L., Hausmann, R. (2007). The Product Space Conditions the Development of Nations. Science, 317(5837), 482–487. Link

  17. 17.

    Traag, V. A., Waltman, L., van Eck, N. J. (2019). From Louvain to Leiden: Guaranteeing Well-Connected Communities. Scientific Reports, 9(1), 5233. Link

  18. 18.

    Philippe Urfalino (2007). La décision par consensus apparent. Nature et propriétés. Revue Européenne des Sciences Sociales, 45(136), 47–70. Link

  19. 19.

    Philippe Urfalino (2014). The Rule of Non-Opposition: Opening Up Decision-Making by Consensus. Journal of Political Philosophy, 22(3), 320–341. Link

  20. 20.

    Ostrom, Elinor (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press. Link

  21. 21.

    Hardin, Garrett (1968). The Tragedy of the Commons. Science, 162(3859), 1243–1248. Link

  22. 22.

    Hathaway, Oona A. (2002). Do Human Rights Treaties Make a Difference?. The Yale Law Journal, 111(8), 1935–2042. Link

  23. 23.

    Hafner-Burton, Emilie M., Tsutsui, Kiyoteru (2005). Human Rights in a Globalizing World: The Paradox of Empty Promises. American Journal of Sociology, 110(5), 1373–1411. Link

  24. 24.

    Gardiner, Natasha Blaize, Gilbert, Neil, Liggett, Daniela, Bode, Michael (2025). Measuring the Performance of Antarctic Treaty Decision-Making. Conservation Biology, 39(1), e14349. Link

  25. 25.

    Flack, Jessica C. (2017). Coarse-graining as a Downward Causation Mechanism. Philosophical Transactions of the Royal Society A, 375(2109), 20160338. Link

  26. 26.

    Bachrach, Peter, Baratz, Morton S. (1962). Two Faces of Power. American Political Science Review, 56(4), 947–952. Link

  27. 27.

    Jones, Bryan D., Baumgartner, Frank R. (2005). The Politics of Attention: How Government Prioritizes Problems. University of Chicago Press

  28. 28.

    van Elteren, Casper, Lippert, Fiona, Carter, Zach, Kleshnina, Maria, Bode, Michael, Vasconcelos, Vítor V. (2026). Attention shapes collective decisions for the Antarctic global commons. Draft manuscript. Link

  29. 29.

    van Elteren, Casper (2026). Decisions at the end of the world. Draft manuscript. Link

  30. 30.

    van Elteren, Casper (2026). Under the rule of non-opposition, agreement and costly objection leave the same record. Draft manuscript

  31. 31.

    van Elteren, Casper, Kleshnina, Maria (2026). How to Agree Without Agreeing. Draft manuscript

  32. 32.

    Antarctic Treaty Consultative Meeting (2022). Final Report of the Forty-fourth Antarctic Treaty Consultative Meeting, Berlin, Germany, 23 May--2 June 2022. Paragraphs 94--99 and 134

  33. 33.

    Antarctic Treaty Consultative Meeting (2024). Final Report of the Forty-sixth Antarctic Treaty Consultative Meeting, Kochi, India, 20--30 May 2024. Paragraphs 118--128

  34. 34.

    Chomsky, Noam (1957). Syntactic Structures. Mouton

  35. 35.

    Hemmings, Alan D. (2018). Liability Postponed: The Failure to Bring Annex VI of the Madrid Protocol into Force. The Polar Journal, 8(2), 315–332. Link

  36. 36.

    Ferrada, Luis V., Caldera, Diego A. (2024). Negotiation, Adoption and Domestic Implementation of the Annex on Liability to the Protocol on Environmental Protection to the Antarctic Treaty. Antarctic Science, 36(5), 419–438. Link

  37. 37.

    Bastmeijer, Kees (2018). Introduction: The Madrid Protocol 1998--2018. The Need to Address `the Success Syndrome'. The Polar Journal, 8(2), 230–240. Link

  38. 38.

    Beck, Peter J. (1986). The International Politics of Antarctica. Croom Helm

  39. 39.

    Yao, Joanne (2022). An International Hierarchy of Science: Conquest, Cooperation, and the 1959 Antarctic Treaty System. European Journal of International Relations, 28(4), 939–963. Link

  40. 40.

    Nuclear Threat Initiative (2024). Antarctic Treaty. Link

  41. 41.

    Joyner, Christopher C. (1998). Governing the Frozen Commons: The Antarctic Regime and Environmental Protection. University of South Carolina Press

  42. 42.

    Hughes, Kevin A., Gray, Andrew D., Ager, Beverley J. (2024). Attainment of Consultative Status by Parties to the Antarctic Treaty: Past, Present and Future. The Polar Journal, 14(2), 560–591. Link

  43. 43.

    Antarctic Treaty Consultative Meeting (2025). Rules of Procedure for the Antarctic Treaty Consultative Meeting and the Committee for Environmental Protection. Link

  44. 44.

    Antarctic Treaty Secretariat (2025). Delegates Manual. ATCM 47, AD001. Link

  45. 45.

    Antarctic Treaty Secretariat (). ATCM and Other Meetings. Link

  46. 46.

    Antarctic Treaty Secretariat (). The Committee for Environmental Protection. Link

  47. 47.

    Antarctic Treaty Consultative Meeting (2003). Measure 1 (2003): Secretariat of the Antarctic Treaty. Adopted at the Twenty-sixth Antarctic Treaty Consultative Meeting, Madrid

  48. 48.

    Antarctic Treaty Secretariat (2025). Final Report of the Forty-seventh Antarctic Treaty Consultative Meeting. Milan, Italy. ATCM paragraph 113 and CEP paragraphs 157 and 251--254. Link

  49. 49.

    Antarctic Treaty Secretariat (2026). Final Report of the Forty-eighth Antarctic Treaty Consultative Meeting. Hiroshima, Japan. ATCM paragraphs 93 and 97. Link

Footnotes

  1. Every 0.1 increase in distance lowers the odds of taking up a concern by 18%. ↩

  2. At a 5/6 chance of another round, longer training brings apparent consensus down to 36.5% of rounds and the share of pairs passing the full check up to 35.3%. ↩

  3. At 23/24, restarting pairs in punishment changes punishment rules for 28.1% of participants, compared with fewer than 0.1% in matched controls restarted in ordinary play. After 2,000 more rounds of learning, apparent consensus is 83.5% in both groups over their final 1,000 rounds. ↩

  4. Among proposals seeking an instrument whose later outcomes we could establish, 97% of revisions and 83% of new proposals yield an instrument within two meetings. Alternative checks leave open whether that later difference is real. ↩

Join the conversation

Comments

Back to Blog

Related Posts

View All Posts »
Who am I when the machines do the work?

The Jetsons imagined a future where machines work and people live. What happens to expertise, authorship, and livelihood when the work really can be handed over?