Abstract
Many official statistics producers, frequently publishing new versions of their data, have their dissemination systems set up to ‘overwrite’ previous versions. While this ‘lean’ dissemination approach is suitable to most official statistics users (as the main user interest is for the most up-to-date versions of official statistics), this poses a problem to certain stakeholders. For instance, if a policy analyst or a researcher carries out an analysis on an official statistics dataset that is subsequently superseded, there will be issues with replicability, since nobody will be able to obtain the exact same results without having access to the data that are no longer being disseminated. Well-known remedies exist (ranging from the analyst saving the data at their end, to the producer adopting a 'full data warehouse' approach by versioning all of their disseminated data) – each of which have their own drawbacks (such as lack of independent verification of authenticity for user-managed solutions, and considerably increased storage and management requirements for a ‘full data warehouse’). In this paper, a solution based on the so called ‘anchoring’ of official statistics data in the EBSI blockchain ledger is presented. It has the capacity to document all disseminated dataset versions of an official statistics producer in a lightweight fashion. As the solution is currently being piloted, the paper comprehensively treats both the user perspective (exactly what analysts need to do to benefit from the authenticable reproducibility being offered) and the producer perspective (including the various technical design choices an official statistics producer would need to make to deploy a blockchain anchoring solution for their own dissemination infrastructure). The paper briefly touches upon other possible blockchain applications for official statistics, as well as EBSI improvements seen as necessary to realise the full potential of blockchain anchoring as a means to achieve authenticable reproducibility.
Get full access to this article
View all access options for this article.
