A database of public domain works could reduce upload filter overblocking; it’s absurd we need one

One of many problems with the upload filters that Article 17 of the EU Copyright Directive will bring in is that they are likely to overblock. That is, they will stop perfectly lawful materials from being uploaded because of flaws in the filters’ algorithms. Among those blocked lawful materials will certainly be public domain items.

The public domain is often merely defined as where copyright materials end up after copyright has expired, or where a creator expressly chooses to place a work without copyright (as with Walled Culture the book). That definition of “not in copyright” is hardly something that can be used reliably in upload filters. Last year, Paul Keller and Felix Reda wrote a paper exploring the issue, and came up with a suggestion for addressing it:

we are proposing that [Online Content Sharing Service Providers] invest in building and maintaining a publicly accessible shared repository containing verified information on openly licensed works and works in the Public Domain.

A quick check by upload filters in such a database could help to stop public domain materials being blocked. The Pirate Party’s MEP Patrick Breyer presented a proposal for the European Union to support a pilot project exploring the idea. The EU Parliament has just approved the funding for the pilot, which would consist of:

a feasibility study, to confirm that there is an actual market failure and to confirm the risk of over-blocking such public domain works, as well as to determine the technical needs, including from platforms, and ensure the buy-in from stakeholders. The project would also develop a prototype database that could be used, referenced and augmented by platforms, content providers, institutions of the GLAM sector (Galleries, Libraries, Archives and Museums) or other non-forprofit organisations working with public domain or freely licensed content. Such public repositories of freely reusable works could help to unlock the societal value of these works, and thereby truly enable access to and promotion of culture, and the access to cultural heritage. The above work is to be seen as a pilot that will develop our knowledge in the field and help confirm the need and opportunity to possibly develop this prototype into a fully fleshed database.

Although it’s good news that practical work towards mitigating the harmful effects of upload filters is being undertaken, it’s also doubly regrettable. First, that the harmful idea of upload filters is an issue at all, after supporters of Article 17 insisted repeatedly – and falsely – that this kind of automated surveillance would not be necessary if the Directive was passed; and secondly, that the public domain is regarded as something weird and exotic, when it should be the baseline against which copyright is measured, and not the other way around.

Featured image produced by Stable Diffusion.

Follow me @glynmoody on Twitter, or Mastodon.