Diversity of attention guide
This page explains how Sharing: Culture and the Economy in the Internet Age uses diversity of attention to compare how audiences distribute interest across many works. The measure sits at the heart of the book's empirical argument that non-market sharing can widen cultural choice rather than concentrate it on blockbusters alone. Return to the Sharing augmented edition homepage for an overview of the full site, or continue below for concepts, data, and related tools.
What diversity of attention measures
When millions of people share or buy cultural works, attention rarely spreads evenly. A few titles dominate charts while a long tail of lesser-known works still attracts some use. Diversity-of-attention statistics summarize that imbalance: they describe how concentrated or spread out audience interest is across a ranked list of works (downloads, queries, sales, or listens). Higher diversity means more works receive meaningful attention relative to the top hits.
Philippe Aigrain applies these measures to file-sharing networks, legal music communities, and commercial channels. The results support the book's claim that peer sharing can surface works that mainstream distribution overlooks. The approach is developed at length in Appendix A: Diversity of attention and discussed throughout the main chapters on measuring use and sustainable financing.
Working with the sample datasets
Three ranked-popularity datasets power the case studies in Sharing. Each is documented with download links and citation requirements on the Datasets page:
- eDonkey server queries (Complex Networks team, UPMC): ten weeks of server activity, reduced to ranked FileId popularity.
- Hungarian BitTorrent movie sharing (Bodó Balázs and Zoltán Lakatos): three ranked distributions covering all files, IMDb-identified files, and per-film compounds.
- Musique Libre / Dogmazic listening and download data: compound-level ranked popularity for a voluntary music community.
Researchers replicating the analysis should credit the original collectors when publishing results. Zip archives and CSV files are hosted alongside brief provenance notes so you can inspect the same inputs the book uses.
Steps to explore the analysis on this site
1. Read the appendix discussion. Start with Appendix A comments to see how readers have interpreted the methods and limitations.
2. Download and inspect the data. Pull the archives from Datasets and verify ranked lists against the book's tables and figures.
3. Review the implementation. Source files referenced in the book are listed on Code. They document how popularity ranks were derived from raw traces.
4. Connect to financing models. Chapter 8 and the economic models page address how policy choices interact with audience diversity. Adjust parameters there to test alternative support schemes for creators.
5. Join the chapter conversation. Questions about measurement choices or policy implications fit well on the chapter comment pages, especially chapters on measuring use and sustainable resources.
Why this matters for policy today
Copyright debates still often assume that uncompensated sharing automatically shrinks cultural variety. The diversity-of-attention evidence in Sharing offers a concrete counter: when sharing is widespread, empirical distributions of attention can be more dispersed than in some commercial channels, not less. That does not settle every legal question, but it reframes them. Policymakers and researchers need metrics that track how many works actually reach audiences, not only how many units sell at the top of the chart.
For the book's broader argument on legitimate non-market sharing and hybrid financing, see the Sharing augmented edition homepage and the open-access text via Download.
