Barcelona, 8–10 September 2026
Digital Statecraft ran as a thread through all three days of Data for Policy 2026 at Universitat Pompeu Fabra. The track was designed to test an argument rather than display it. Over a plenary consultation, two structured debates and 27 presentations across nine parallel sessions, the case for a distinct approach to governing in the algorithmic age was put in front of people who would have to make it work.
They came from statistics offices, ministries, universities, archives and civil society, and they brought systems that are already running: an early-warning tool in English housing services, industrial land allotment in Goa, customs risk scoring at the European Commission, a national geoportal in Uzbekistan, data centre ownership in Australia, the official record in Cameroon, statistical production across Latin America and the Caribbean.
The Digital Statecraft Academy (DSA) holds two conditions together and treats neither as sufficient on its own: the ability to govern in the algorithmic age, and the legitimacy to do so. Across cases with little else in common, the same question kept returning. Principles are now broadly agreed; what has to change inside an institution for them to hold at the moment a decision is made?
Plenary Session: consulting on the Digital Statecraft Manifesto 2026
A year on from DSA’s founding manifesto, the DSA has set out Ten Principles, drawn from those two conditions, to guide the design, implementation and development of the twenty-first-century state. The Manifesto puts the stakes of the pairing plainly. Ability without legitimacy risks oppression: a state that can inspect, audit and enforce, but cannot be challenged over how it does so, has built efficient authoritarianism. Legitimacy without ability is accountable impotence: a state with every democratic mandate and no capacity to understand the systems shaping its citizens’ lives has power on paper only.
The plenary was run as a consultation rather than a presentation. Chaired by DSA co-founders Dr Zeynep Engin and Tim Gordon, the session brought together Dr Viviana Bastidas (University of Cambridge), Dr David Uriel Socol de la Osa (Hitotsubashi University and Oxford University), Prof Irene Ng (Innovorsa) and Dr Stefaan Verhulst (The GovLab, New York University, and The Data Tank).
The Digital Statecraft Manifesto 2026 was launched at the conference. It remains a living document, intended to evolve through collective engagement and real-world testing, as v1.0 did before it. Read and Endorse the Digital Statecraft Manifesto 2026.
Two DS debates
Does the state remain the primary institutional form of governance in the algorithmic age? Prof Irene Ng and Dr David Uriel Socol de la Osa took the two positions, chaired by Tim Gordon. Both began from the state as a hard-won achievement. The disagreement was about what follows: whether digital interdependence makes the state’s coordinating role more necessary than ever, or whether legitimate governance now needs a broader architecture in which the state is indispensable but no longer paramount.
Can governance operate at algorithmic speed? Prof Jennifer Schooling and Dr Cristina Luca (Anglia Ruskin University) argued it out, chaired by Dr Stefaan Verhulst. One case rested on deliberation, judgement and democratic legitimacy, with AI as a tool for reaching goals rather than setting them. The other held that algorithms already perform governance functions at speeds humans cannot match, and that humans may be better placed setting objectives while algorithms govern other algorithms.
Adaptive institutions
Dr Viviana Bastidas (University of Cambridge) and Prof Jennifer Schooling asked what architectures make governance traceable, using an AI-enabled early-warning system in housing services, and found that frontline conditions require designing for auditability and answerability rather than simply keeping a human in the loop. Sem Nouws and Marijn Janssen (TU Delft) proposed boundary objects to close the translation gap between citizens, politicians and experts. Aziza Umarova (Agency for Strategic Development and Reforms) presented Uzbekistan’s national geoportal and a model of outcome-oriented spatial governance that moves from whether a target was met to where progress is happening and why.
Case studies across jurisdictions
Kartik Akileswaran (Growth Teams), documented reform at the Goa Industrial Development Corporation, where the sequence was the point: outreach across the industrial estates first, a single replacement regulatory framework second, the digital platform third. Processing time for transfer of leasehold rights fell from 223 days to 24, and the module still running on unreformed planning rules produced the most friction in the system. Tala Bautista (Kalinga State University) presented Digital Arang, a provincial attempt in the northern Philippines to write AI governance into local law on the authority of Indigenous Knowledge Systems and Practices, and argued that the decisions a community will not delegate cannot be fixed as a list of prohibited uses. They have to sit in customary deliberation itself, with autonomy understood as the capacity to set terms of engagement rather than as self-sufficiency.
State and citizenship
Dr Sarah Barns (RMIT University) presented a facility-level audit of Australia’s data centre ownership and argued for sovereignty understood as relational practice. Prof Irene Ng (Innovorsa) and Susan Wakenshaw (University of Warwick) argued that subject-generated data now functions as economic identity infrastructure, which data protection regulates without reconstituting who controls it. Mohammed Alotaibi (UNU-CRIS) developed cognitive democracy, treating the state as an intelligence system that must sense, diagnose, coordinate, remember and adapt. Dr David Uriel Socol de la Osa defined sovereignty as the capacity to act under interdependence rather than as technological autonomy.
Digital infrastructure
Dr Mireia Paulo Noguera (Ruhr-Universität Bochum) traced China’s AI governance from 2012 to 2026 as an extension of an existing securitised data regime. Brigitta Németh and Johannes Wachs (Corvinus University of Budapest) studied domestic platform-building through Gitee, asking how dependence on global open source can be reorganised rather than eliminated.
Governing at the margins
Manlin Li, Prof Jennifer Schooling (Anglia Ruskin University) and Panpan Sun (University of Nottingham) identified an accountability mismatch in UK digital governance: exclusion is structural and collective, while remedies are fragmented and individual. Prof Victoria Lemieux (University of British Columbia) presented Clio-X and Cameroon’s Official Gazette, treating archives as upstream of AI governance rather than downstream. Dunja Bonacci Skenderović showed how Croatia’s administrative data on violence against women counts cases without revealing patterns.
AI and compliance
Halfdan Lynge (Danish Agency for Digital Government) tested Denmark’s metadata catalogue against the demands of AI agents, and argued that catalogues built as instruments for human users now need to work as infrastructure. Andrea Fernandez Conde (INEGI) reported a sharp rise in AI-origin traffic to Mexico’s statistics portal alongside falling human visits, and set out the layered architecture INEGI is building so that agents interpret official statistics correctly. Andrea Cerasa (European Commission, Joint Research Centre) presented a reliability score for EU customs risk management that tells officials when a benchmark price can be trusted, without replacing their judgement.
Governability
Kai-Hsin Hung (HEC Montréal) and co-authors mapped 46 AI infrastructure investments in Africa totalling USD 12.7 billion, finding that control stays concentrated in the compute layer: hosting data centres is not the same as governing them. Fortunato Musella, Luigi Rullo and Sara Concetta Santoriello (University of Naples Federico II) offered a digital policy design framework as a diagnostic tool for the Ten Principles. Mario Alejandro Tabares Arango and Mateo Maya (CORLIDE, Medellín) named zombie governance, where adopted policies stay disconnected from the decisions they govern, and proposed a shared semantic layer as the bridge. Hugo Leal (University of Cambridge) and R. Marie Santini (NetLab UFRJ) set out what restricted access to platform data costs evidence-based policymaking.
Legitimacy in algorithmic mediation
Anna Nesvijevskaia (HES-SO) asked who is legitimate to define suspicion when fraud evolves faster than the institutions that qualify it. Liat Hasenfratz, (Princeton University) argued that institutions must govern not only access to an archive but what it may make knowable over time. Hon Bor So examined 46 high-stakes UK algorithmic transparency records, distinguishing system-level visibility from the ability to reconstruct a specific decision afterwards.
Authority and accountability
Rick Dane Moore (White Buffalo Alliance) proposed governance public infrastructure that sequences rights, duties, consent and remedy before computation, drawing on Indigenous constitutional practice. Dr Luca Di Gennaro Splendore (University of Malta) mapped who actually produces official statistics across Latin America and the Caribbean, finding central banks in a central role and democratic accountability shaping the results. Maria Lorena Florez Rojas (University of Groningen) separated accountability, responsibility and liability in EU public administration.
What the track showed
The Manifesto rests on a claim: each Principle names a requirement whose absence produces a specific, identifiable governance failure. The three days in Barcelona worked as a field test of it, and the failures described were not hypothetical.
Authority was visible at one point in a chain and lost at another, whether in UK transparency records, African compute or outsourced model evaluation (Principles 04 and 08). Rules aged faster than the systems they covered, or never reached the decision they were meant to govern (03 and 06). Exclusion was structural while the remedies on offer stayed individual (07). Archives and administrative data quietly settled what could be known later, and by whom (09). The most careful work kept the human inside the decision rather than beside it (02 and 10).
The debates held the same tension open at a higher level. The state may remain primary because it alone currently holds the full set of legitimacy conditions, and governing may resist algorithmic speed, but the systems already running at that speed still have to be governed by something. Neither question closed, and neither was meant to.
The gap now sits between what institutions have declared and what they can do at the moment a decision is made. That is the ground the Digital Statecraft Manifesto 2026 sets out to cover, and the reason it is published as a living document rather than a finished one. Endorsements can be added through the Manifesto page, and enquiries are welcome at office@digitalstatecraft.academy.
Learn more about the DS Special track here.