Global analysis
How autocracy shapes the social sciences
Are the contents and conclusions of the social sciences different in autocracies and democracies?
ERC Consolidator Grant · University of Oslo
How authoritarian rule shapes what we know
AutoKnow asks whether science is affected by authoritarian government, and if so, how. The project develops new theories of science under authoritarian rule and tests these with large-scale bibliometric data, surveys, case studies, and computational text analyses of what scientists actually publish.
Principal investigatorTore Wig
HostUniversity of Oslo
FundingERC Consolidator Grant
Period2025 – 2030
The puzzle
Science is supposed to need freedom. Autocracies are becoming science leaders.
The growth of scientific knowledge is one of the towering achievements of humankind, yet we still understand little about what makes some societies more fertile for science than others. A long tradition holds that science requires protected freedom of expression and open inquiry. At the same time, authoritarian regimes such as China, Russia, and Saudi Arabia have massively increased their investment in science, and the international scientific frontier is increasingly traversed by countries with unfree elections, weak civil liberties, and limited academic freedom.
Authoritarian rule is not confined to autocracies. Authoritarian leaders and movements have become a major political force inside democracies, where they attack universities, scientists, and scientific institutions. Understanding how authoritarian rule affects science is therefore vital to the future of scientific progress itself, and to the challenges that depend on it, from global health and climate change to governance and AI.
Is science affected by authoritarian government, and if so, how?
AutoKnow answers this science–authoritarianism puzzle by building the first comprehensive social-scientific account of how authoritarian regimes, and authoritarian leaders in backsliding democracies, constrain scientists and shape the knowledge they produce.
Beyond counting papers
What we measure
Conventional bibliometrics count publications and citations. AutoKnow starts there, but its core is a set of deeper, content-based dimensions of scientific output, measured with large language models and computational text analysis across the period 1950–2025.
01
Volume and impact
Publication counts and citations across and within disciplines, the workhorse metrics of country-level scientific performance.
02
Scientific progress
Whether published work disrupts established knowledge, contributes novel ideas, and takes intellectual risks.
03
Political bias
Whether scholarship on the social world tilts in favour of, or against, governments, regimes, and political authority.
04
Censorship and self-censorship
Conspicuous absences: topics, concepts, and findings that should appear in a literature but do not.
Research questions
Four questions
- How do authoritarian regimes and leaders attack and constrain science?
- What is the effect of authoritarian rule on the volume, quality, and content of scientific output?
- How do individual scientists adapt, resist, or migrate?
- What are the downstream effects on the global stock of knowledge?
Approach
Big data, new measurements
Global bibliometric data
Millions of peer-reviewed articles from the Web of Science and OpenAlex, matched to regime data from V-Dem, from 1950 to the present.
Attacks on science
A new dataset of attacks on scientists, universities, and scientific institutions by authoritarian regimes and leaders.
Surveys of scientists
Original survey data on scientific norms, self-censorship, and how researchers adapt to political pressure.
Language models at scale
LLM-based classification and multi-agent analysis pipelines that read abstracts across languages and measure framing, sensitivity, and bias.
AI, transparently. Language models are part of the project's methods and AI assistants are part of its workflow. Both are governed by a written policy on logging, disclosure, and human accountability. Read the principles.
Research in progress
Current papers
A growing set of papers is currently in production, alongside a planned monograph.
Methods
Signals of autocratic effects on scientific work
How do autocratic regimes shape what scientists write?
Content analysis
Autocracy and the production of historical knowledge
Does authoritarian rule shape what historians write about their own nation's history?
Case study
Social science in autocracies: evidence from the GDR
What did four decades of authoritarian rule do to social science in East Germany?
Case study
Academic output under democratic backsliding: evidence from Turkey
How has autocratic consolidation in Turkey changed what academics study and how they write?
Survey experiment
Political constraints on scientific credibility
Does the credibility of scientists depend on their politics, and on the regime they work under?
Theory
The regime model-alignment problem
What are the costs and benefits, for democracies and autocracies, of aligning AI models with the regime?
Global analysis
The political origins of critical social science
Where does critical social science come from, and did the student protest movements of the 1960s and 1970s leave a lasting imprint on it?
Attacks
Does academic output predict attacks on scientists?
Can what institutions publish predict which of them come under academic-freedom attacks?
Methods
Agentic architecture for evaluating theories of politics
How can the evidence for a "grand theory" with many implications be weighed systematically?
Theory Lab
Autocracy and science. Sharpening theories using an automated theory-lab approach
Of the many plausible theories of how autocracy shapes science, how do we decide which ones to test?
People
Team
AutoKnow is led by Tore Wig, Professor of Political Science at the University of Oslo, together with associated researchers at Oslo and Stavanger and co-authors in Berlin, Dublin, Konstanz, and Santiago. The project is recruiting two postdoctoral fellows and two PhD fellows.
Join the project. Interested in working on science, autocracy, and computational social science? See open positions.
Updates
Latest
Project website launched with an overview of the growing set of papers currently in development.