The French parliament is currently examining a significant law on « The Digital Republic ». It covers a wide array of topics: open data, open access, Freedom of panorama, data portability, digital platform regulation… Contrary to the initial plan of the French governement, several MPs have put forward another item to this list: a text and data mining exception. While approved by the French lower house, the final fate of the exception is still uncertain at this stage. The governement initially wanted to set a second vote, that has finally been upheld, so that the mining exception currently figures prominently in the « Digital Republic » law. Anyhow, it still has to gain the approval of the senate.
Mining and its legal issues
The current intellectual property regulation in Europe restrains an extraordinary scientific opportunity: being able to grasp the main features, ideas and entities of millions of texts. Both massive digitization of corpora and sophistication of mining and clustering tools allow to lead « distant reading » on an unprecedent scale. The Text2genome project can draw an immense state of the art from genomic literature; ContentMine ambitions to liberate 100 millions facts from academic publications. I’m both an advocate and a pratician of text and data mining. My main focus concerns 19th century generalist and scientific periodicals — which are fortunately in the public domain, but I would be very happy to extend my scope beyond the beginning of the XXth century.
Yet, mining necessarily implies to reproduce: mostly for performance issues, online text or data has to put in a « local » database (it would be highly limitative to scrape on the fly). Within research projects, this output has to be shared to all the active members. As such, mining even lawful ressources infringes intellectual property rights. As absurd as it may sounds, the right to read does not entail the right to mine (which is nevertheless nothing more than an algorithm-assisted reading…).
New policies try correct this discrepancy. The United Kingdom has passed a text and data mining exception in 2014. This is a very hot topic within the ongoing European reform of intellectual property right. And now, come France.
The terms of the French Exception
The French exception reflects a consensual demand. All the major tendencies from the communist to the main right-wing party, the Republican, has proposed a similar amendment. I’ve made a very rough translation — certainly not an official legal translation by any means, but still better than Google Translate:
The Code of Intellectual property is modified as such:
1. After the 9th paragraph of the L. 122-5 article, a 10th paragraph is added:
« 10. The digital copies of reproductions made from a lawfull source, in order to mine text and data for the need of public research, all commercial end being excluded. A decree define the conditions by which the mining of text and data can be performed, as well as the modalities of preservation and communication the output files at the end of the research process ; theses files form research data. »
2. After the 4th paragraph of the L. 342-4, a 5th paragraph is added:
« 5° The digital copies of reproduction made by a person who has a lawfull access, in order to mine text and data in a research aim, all commercial ends being excluded. The preservation and communication of the technical copies from the data processings, at the end of the research activities for which they have been produced, are performed by certified organisations named by decree. The other copies or reproductions are deleted. »
The two articles add another element to previous list of exceptions to the French Intellectual property code. In European countries, databases are actually covered by two different rights: authorship rights (as possible « original » compilation of data) and « sui generis » rights (a special status given to the producer of the database, that gives him the right to contest any substantial copy of the database — as a side note, this right is currently under heavy criticism from the European Parliament, that is contemplating its complete deletion). The first articles focuses on the exceptions to authorship rights ; the second one targets the exceptions to « sui generis » rights.
The French exception is roughly similar than its British counterpart. They are both limited to scientific research ; they both aim to link the right to read and the right to mine, by giving a license to extract to researchers who have a lawfull access to a resource. There is nevertheless an additional restriction regarding the preservation of the output. So far the British exception does not seem to set any temporal limitation to the keeping of the « local » databases.
A first draft of the French exception actually envisioned to have the output fully deleted once the research was over. This would have proved quite unpractical. As every scientific work, text and data mining project are perfectible: the output is so large that it can be of use for many unexcepted outlooks. Fortunately, the adopted amendment comes to a better term: the output would be kept by a « certified » organisations (the French National Library is very likely to be one of them), and could therefore be reused.
A long road to go…
I have followed the making of the law from quite some time. Two years ago, the French Literary, Artistic and Scientific Council for Intellectual Property (CSPLA) initiated a consultation on text mining. I represented the collective Savoirscom1 with Lionel Maurel. We were the only member of « civil society » heared: most of the interlocutors were publishing companies (such as Elsevier or Springer, which concurely organized a « Licenses to Europe » process in Bruxelles with the final aim of transforming data mining as an additionnal service, fully controlled by the publishers). We argued in favor an exception on the ground of a « fair dealing » limited to scientific use (in a very similar way to the envisionned british exception, that was not adopted at the time). This measure was to be completed by a positive definition of the « informational public domain » in order to avoid the excessive extension of « sui generis » rights toward « raw » facts and data (cf. the synthesis of our argument).
Logically enough, our solitary position did not favor our arguments. The final report, publised by the end of 2014 strongly advised against an exception. Only, by this time, the debate has shifted. The debate surrounding the Elsevier National License — a 172 millions euros contract between the world leading publisher and French Universities that I leaked with Rayna Stamboliyska — gave a significant boost to the idea of pro-open access regulations.
While this has not been initially planned, the main law passed by the socialist legislature on digital issues includes an « open access law ». In a similar pattern to recent german and italian exceptions, it confers to researchers the ability to republish their work regardless of the terms of the contract they sign with their publisher. A mining exception was added to the package in a draft version of the law in july 2015… and finally removed in the first public version in september. It was a timid thing: researchers were forced to delete the output at the end.
In-between, this technical measure, that merely relieved scientist of useless intellectual property barrier, has become a moving target of the rightholders lobbies. A counsellor of the French National Publishers Syndicate (Syndicat national des éditeurs) issued an angry brochure against intellectual property exceptions and, more widely, against immaterial commons: Free price is stealing (La gratuité c’est le vol). Ironically enough, the brochure was distributed free of charge. The mining exception envisionned in the European reform of author rights and in the French « Digital Republic » law was hold as some kind of communist apocalypse:
If such investments could be lawfully plundered, no publishers would ever grant any investment for digital tools. There is, in reality, no economic activity in the world of which the productions can be freely extorted for private use and without any compensation.
Yet, the fight was not unequal. An extraordinary procedure gave scientific communities the ability to speak up in favor of better terms for the open access law and in favor of the reintroduction of mining exception: apparently for the first time in France, a law was examined, evaluated and completed on an online platform. The « République Numérique » website hosted several thousand contributions from virtually everyone: lobbies, companies, public institutions, communities and simple citizens… Several of them argued in favor of a mining exceptions, and they were extremely well received.
This open consultation has given a lot of additional momentum to innovative measures that would have been otherwise upheld by powerful lobbies, or simply not conceived in the first place. MPs seem to have been particularly sensitive to original perspectives. Full Freedom of Panorama (that includes commercial use) was actively supported by one fifth of the lower house (more than one hundred MPs actually signed at least one favorable amendement). While MPs were more open-minded, scientific communities proved bolder : a few days ago, the President of Universities Collective (CPU) and the National Centre for Scientific Research (CNRS) signed a strong committment toward an open access law and a mining exception.
Rigthholders lobbies are still far from powerless. They were able to have a significant innovation partly removed from the law: a positive recognition of « informationnal commons » (a wider notion that cover both the public domain of « works », the public domain of « informations » and voluntary Free knowledge such as Wikipedia). Nevertheless, they have never been in such a weak position in the recent past.
All in all, the game remains fully open. The current state of the « open access » law reflect this tensed debate between coalitions of actors of gradually equal strength and influence: while scientific institutions, public libraries and open knowledge communities were able to append data to the focus of the law (something their german counterpart has not succeeded in) the publishers have, no less successfully, vetoed the inclusion of collected conference contributions. The suspense should still last till the end of the legislative procedure (probably no more than a month).