Periodic Reporting for period 1 - DALOSS (Data loss: the politics of disappearance, destruction and dispossession in digital societies)
Período documentado: 2023-06-01 hasta 2025-11-30
Digital technologies reshape how societies store and preserve information. While we focus on vast data creation, an equally important phenomenon occurs: data constantly disappears. From broken website links to deleted social media posts, digital information vanishes at unprecedented rates. This data loss is not merely technical failure—it reflects political and social dynamics about what gets preserved in our digital age.
The DALOSS project investigates data loss as governance, examining how decisions about digital information shape power relations in contemporary societies. Rather than treating data loss as unfortunate side effects, the project reveals it as active processes through which institutions, platforms, and governments manage information and control narratives.
The project addresses critical knowledge gaps. While extensive research examines data collection and surveillance, much less attention focuses on deletion and forgetting mechanisms. Most studies treat data loss as either technical failure or deliberate censorship, missing complex governance processes between these extremes. Existing approaches rarely connect data loss practices across institutional contexts—from government archives to social media platforms to artificial intelligence systems.
Expected impact extends beyond academic knowledge to influence digital preservation practices, inform policy debates about digital rights, and support communities maintaining their digital heritage. By understanding how data loss operates as governance, societies can make informed decisions about digital preservation and protect vulnerable communities from digital erasure.
The project integrates social sciences and humanities perspectives essential for understanding data loss as cultural and political phenomenon. While computer science focuses on technical solutions, humanities and social science methods reveal power relations, institutional practices, and cultural meanings in data loss processes. This interdisciplinary approach connects technical analysis with critical examination of whose voices get preserved and whose disappear from digital records.
The DALOSS project investigates data loss as governance, examining how decisions about digital information shape power relations in contemporary societies. Rather than treating data loss as unfortunate side effects, the project reveals it as active processes through which institutions, platforms, and governments manage information and control narratives.
The project addresses critical knowledge gaps. While extensive research examines data collection and surveillance, much less attention focuses on deletion and forgetting mechanisms. Most studies treat data loss as either technical failure or deliberate censorship, missing complex governance processes between these extremes. Existing approaches rarely connect data loss practices across institutional contexts—from government archives to social media platforms to artificial intelligence systems.
Expected impact extends beyond academic knowledge to influence digital preservation practices, inform policy debates about digital rights, and support communities maintaining their digital heritage. By understanding how data loss operates as governance, societies can make informed decisions about digital preservation and protect vulnerable communities from digital erasure.
The project integrates social sciences and humanities perspectives essential for understanding data loss as cultural and political phenomenon. While computer science focuses on technical solutions, humanities and social science methods reveal power relations, institutional practices, and cultural meanings in data loss processes. This interdisciplinary approach connects technical analysis with critical examination of whose voices get preserved and whose disappear from digital records.
The DALOSS team conducted extensive fieldwork across multiple institutional contexts to understand how data loss operates in practice. Research activities include ethnographic studies in Danish national and municipal archives, technical analysis of artificial intelligence platforms, and computational investigation of web archive loss patterns.
In artificial intelligence research, the team developed innovative methods for studying closed platform environments where traditional access is impossible. By analyzing technical objects like deprecated datasets and algorithmic interfaces alongside developer documentation, researchers revealed how AI systems manage data preservation and deletion. Focus on emerging "machine unlearning" techniques shows that AI forgetting involves removing attention from content rather than deleting entirely, revealing complex politics of knowledge in algorithmic systems.
Archive research examined traditional preservation institutions and emerging digital preservation challenges. Fieldwork in Danish institutions revealed how municipal employees navigate uncertainty about digital infrastructure when making deletion decisions. Research investigated cutting-edge preservation technologies including silver halide microfilm, ceramic inscription, and silicon engraving for long-term storage. This work shows how evolutionary metaphors naturalize certain data loss while obscuring power relations in preservation decisions.
Web archive analysis developed computational methods for studying link rot across multiple archival infrastructures. Using a pilot dataset of over 200,000 data points, researchers mapped systematic patterns of web content disappearance, establishing protocols for large-scale analysis comparing how different archival institutions produce varying loss patterns.
The team created a tripartite analytical framework organizing data loss research around interfaces (how loss appears to users), ecologies (institutional relationships), and storage (material infrastructures). This framework enables researchers to connect experiential dimensions of data loss with underlying technical and political-economic structures.
Collaborative partnerships with institutions including Internet Archive, Danish National Archives, and Bodleian Libraries provided research access and opportunities to influence preservation practices. Public engagement included high-profile media contributions addressing policy implications of data loss during political transitions.
In artificial intelligence research, the team developed innovative methods for studying closed platform environments where traditional access is impossible. By analyzing technical objects like deprecated datasets and algorithmic interfaces alongside developer documentation, researchers revealed how AI systems manage data preservation and deletion. Focus on emerging "machine unlearning" techniques shows that AI forgetting involves removing attention from content rather than deleting entirely, revealing complex politics of knowledge in algorithmic systems.
Archive research examined traditional preservation institutions and emerging digital preservation challenges. Fieldwork in Danish institutions revealed how municipal employees navigate uncertainty about digital infrastructure when making deletion decisions. Research investigated cutting-edge preservation technologies including silver halide microfilm, ceramic inscription, and silicon engraving for long-term storage. This work shows how evolutionary metaphors naturalize certain data loss while obscuring power relations in preservation decisions.
Web archive analysis developed computational methods for studying link rot across multiple archival infrastructures. Using a pilot dataset of over 200,000 data points, researchers mapped systematic patterns of web content disappearance, establishing protocols for large-scale analysis comparing how different archival institutions produce varying loss patterns.
The team created a tripartite analytical framework organizing data loss research around interfaces (how loss appears to users), ecologies (institutional relationships), and storage (material infrastructures). This framework enables researchers to connect experiential dimensions of data loss with underlying technical and political-economic structures.
Collaborative partnerships with institutions including Internet Archive, Danish National Archives, and Bodleian Libraries provided research access and opportunities to influence preservation practices. Public engagement included high-profile media contributions addressing policy implications of data loss during political transitions.
The DALOSS project advances understanding of data loss from purely technical problem to complex governance mechanism. This reframing enables new approaches to digital preservation, platform regulation, and information rights that account for political dimensions of technological systems.
Key innovations include developing research methods for studying closed digital platforms, creating computational techniques for systematic web loss analysis, and establishing theoretical frameworks connecting data loss across institutional contexts. The interdisciplinary approach bridges computer science, archival studies, and critical social theory revealing previously invisible aspects of digital governance.
Machine unlearning research provides crucial insights for AI regulation and ethics. As AI systems become central to social decision-making, understanding how they selectively "forget" information affects accountability, bias correction, and rights to be forgotten. Findings that unlearning operates through attention rather than deletion suggest current regulatory approaches may be insufficient.
Web archive research establishes foundations for systematic analysis of digital cultural heritage preservation. Developed methods enable comparative analysis of how different preservation approaches affect what gets saved for future generations, supporting evidence-based digital preservation policy.
The analytical framework provides tools for researchers, policymakers, and technologists across diverse digital contexts. By organizing data loss analysis around interfaces, ecologies, and storage, the framework enables systematic investigation of deletion politics from user experience through institutional practices to material infrastructures.
Future research includes scaling web loss analysis to comprehensive international datasets, extending AI forgetting research to emerging language models, and developing community-centered approaches to digital preservation. Policy applications include informing platform regulation accounting for deletion alongside data collection and developing institutional practices that democratize decisions about digital memory.
The collaborative approach demonstrates how academic research can contribute to preservation communities while advancing scientific knowledge, creating sustainable knowledge exchange supporting both research excellence and practical digital heritage preservation.
Key innovations include developing research methods for studying closed digital platforms, creating computational techniques for systematic web loss analysis, and establishing theoretical frameworks connecting data loss across institutional contexts. The interdisciplinary approach bridges computer science, archival studies, and critical social theory revealing previously invisible aspects of digital governance.
Machine unlearning research provides crucial insights for AI regulation and ethics. As AI systems become central to social decision-making, understanding how they selectively "forget" information affects accountability, bias correction, and rights to be forgotten. Findings that unlearning operates through attention rather than deletion suggest current regulatory approaches may be insufficient.
Web archive research establishes foundations for systematic analysis of digital cultural heritage preservation. Developed methods enable comparative analysis of how different preservation approaches affect what gets saved for future generations, supporting evidence-based digital preservation policy.
The analytical framework provides tools for researchers, policymakers, and technologists across diverse digital contexts. By organizing data loss analysis around interfaces, ecologies, and storage, the framework enables systematic investigation of deletion politics from user experience through institutional practices to material infrastructures.
Future research includes scaling web loss analysis to comprehensive international datasets, extending AI forgetting research to emerging language models, and developing community-centered approaches to digital preservation. Policy applications include informing platform regulation accounting for deletion alongside data collection and developing institutional practices that democratize decisions about digital memory.
The collaborative approach demonstrates how academic research can contribute to preservation communities while advancing scientific knowledge, creating sustainable knowledge exchange supporting both research excellence and practical digital heritage preservation.