White Paper — Open Science in a Digital Republic|
Contribution of the CNRS Scientific Board
1The place of digital technologies in scientific activity has today become capital, although of course, it is important to realise that they only partly freeze or take a “snapshot” of reality and research in a given state. Scientific activity has many other facets than the management of data. However, the digitisation of the data used by scientists and their publications enables automated processing, fast transfer, the harmonisation of access methods and descriptions; all these advantages help bring vast, rich and diverse resources within the reach of researchers, with much shorter lead times. By releasing the scientists from certain repetitive and time-consuming tasks, digital technology can therefore free up their reflexive and creative abilities.
2As has been written many times, it is no doubt possible in this regard to compare the opportunities made available to research by digital technologies with those familiar to the scholars of the sixteenth and seventeenth centuries, with the invention of the printing press and the resulting acceleration in exchanges of knowledge.
3In the human and social sciences, with regard to recent scientific publications, while more and more French-language journals offer free access immediately or after a few years (mainly through HumaNum, BSN, OpenEdition), English-language journals are often confined to rather expensive platforms; the parallel submission of articles to open archives is relatively undeveloped. The scanning of printed sources used by a number of disciplines, whether it relates to the oldest scientific publications, or novels, journals, legal treaties, etc. is well on track; this is often available through open access, although some large companies (such as Gale) also produce databases at prohibitive prices, which are virtually inaccessible in Europe.
4Platforms for exchanging numerical data, whether this relates to the most detailed scales of official statistics or data produced by research, have also been established (Quetelet Network, DIMESHS, etc.): they provide better circulation of data, compliance with the necessary constraints, such as anonymisation, and the documentation (“metadata”) without which the figures would be unusable.
5That said, the data from the human and social sciences, which vary greatly according to the disciplines (from history of art to economics, and including linguistics), are far from being limited to copyright-free printed materials or figures. Platforms for sharing still need to be created, for example, for photographs from archives or photos of works of art taken with a scientific purpose (which raises the issue of the right of reproduction); they are still relatively undeveloped for data from qualitative field surveys (which pose complex problems of anonymisation, formatting and documentation). The problem here is that some of the data used by scientists in the HSS were not produced by them (this may concern a song, a company’s annual report or the architecture of a monument): other natural or legal entities have rights over them. Data sharing and “data- and text-mining” techniques are thus unevenly spread depending on the types of data, mainly due to legal obstacles and a lack of human resources for the production and maintenance of quality metadata. Due to these constraints, for many types of data in the HSS it seems difficult to imagine free sharing that would go beyond sharing for scientific use, with all the difficulties presented by the definition of this scope. Besides, there would indeed be a danger of appropriating data that may be highly sensitive. Moreover, for some types of data, the exploitation period before publication may be rather long, which is an argument in favour of embargo periods before they can be shared being adjusted to take these specific characteristics into account.
6In astronomy, and more generally in areas of the science of the universe or science of observation, the paradigm of the virtual observatory is becoming widespread. The data are freely accessible in astronomy for the entire community after expiry of a proprietary period. The preferred approach is the maximised reuse of data. To achieve this, the formats, descriptions and modes of access to archive data, metadata and the applications likely to be used to process them should be harmonised and standardised, in order to achieve interoperability. This interoperability extends to the linking of research data with online publications. The dangers of appropriation for commercial purposes have not been very pressing up to now, although things could change in the future (for example with space meteorology and the detailed observation of solar eruptions).
7In biology, digital publishing is widespread and academic institutions have developed platforms to help researchers find articles, and access to abstracts is free. The most important (PubMed) is offered by the US National Institutes of Health (NIH). Access to all the articles is generally for a fee, with transfer of copyright to the publisher being the most common practice. It should be noted that the NIH has objected to this practice and proposes open access, via PubMed, to an unformatted version of any article published by a publisher describing work funded by the NIH. Open access has been developing over the past decade. The cost of publication is then generally paid by the authors on publication.
8While text-mining techniques are not a priority for most fields of biology as a discovery tool (but rather in terms of documentary collection), data mining itself is playing an increasingly important role. Free access to these data is widespread, as happened concerning the human genome. Many publishers, including Nature, also make publication of an article conditional on the depositing of mass data associated with a publication on a platform that is accessible to all, free of charge. It should be noted that this requirement goes beyond digital data and also concerns material produced within the framework of the corresponding research.
9When a paper describes a particular material (cell lineage, microorganisms or genetically modified mice, virus, antibodies, etc.), the publisher (Nature, etc.) asks the author to commit to donating this material to other academic researchers. International platforms exist for the storage and distribution of this material. Beyond the question of mass data, several publishers including Nature are considering implementing a system allowing access, via their sites, to the raw data that led to the development of the figures from an article. While this will help the reader ensure the correct interpretation of the results, the question arises of the ownership of these data and their eventual transfer.
10With regard to physics apart from “major instruments”, open access to raw data is not yet very widespread. In contrast, many digital libraries have been formed and made freely accessible by groups of researchers; regularly updated, they relate as much to the theoretical modelling of generic problems (electrical conduction, molecular dynamics) as to the development and management of experiments (interfacing of devices, libraries for processing data). Digital technologies also play an essential role in the dissemination of results, with the almost systematic use of pre-publication servers. Articles are deposited on these servers at the same time as they are sent to a peer-reviewed scientific journal, enabling readers to take early notice.
11The field of chemistry is in fact really a bridge between the practices of the life sciences and those of physics. The rule is publications in paid journals from learned societies (American Chemical Society, Royal Society) or commercial companies (Wiley, Elsevier, etc.) and the timid development of “gold-”type open access, paid for by the authors. There are in fact few differences between the two; negotiations with the ACS were for a time harder than with Elsevier. There is no pre-publication archive like ArXiv. Freely accessible databases are developing, especially the Cambridge Structural Database that contains all the published molecular structures.
12In mathematics, databases relating to publications are very important for both individual and community work. A unique feature of this discipline is the importance of easy access to “old” publications (i.e. several years, decades or even centuries old). Long-term access to these publications is therefore crucial for research. Publication archiving platforms such as HAL or ArXiv thus respond in part to this problem and should be supported, along with metadata platforms (MathSciNet, Zentralblatt, etc.). As regards digital data, for issues of reproducibility, and comparison and interpretation of simulation and calculation methods, they need to be freely accessible and also maintained in the long term (archival, catalogues of datasets, etc.), and this concerns both software and computing code. Furthermore, mathematics plays an important role in the analysis, management and exploitation of masses of data (the issues surrounding Big Data). It is certainly very important for the data to be accessible, but when they become more and more massive, it must also be possible to exploit them effectively. In this area there are important challenges to be addressed for mathematical research.
13An important question spanning all the disciplines is that of the “publication” of data. The requirement for free access is clear in the case of data associated with publications that have been duly validated by peer-reviewed journals. But what about data that may be placed online before publication, for example for analysis and interpretation as part of a broad collaboration? This is a growing reality in a number of disciplines. This problem is especially acute since the definition of what constitutes published data is sometimes vague.
14It can therefore be seen that, as with any type of progress, the digitisation of data and scientific results can have counter-effects. Scientific results play a key role in global economic competition by conferring sometimes considerable competitive advantages on their holders. In return, in order to develop and experiment, modern science needs the technologies often supplied by the world of production, which is governed by market forces. This is particularly true of the features that can be provided by scientific publishers.
15Yet it is universally recognised that knowledge development occurs through the exchange of ideas, results and data between scientists. It is therefore vital to limit the appropriation of scientists’ work by private interests and at the same time to provide a legal framework to free up as far as possible the exchange of data for scientific use.
16The Scientific Board supports three important principles that would meet these objectives:
The complete freedom of circulation and use of scientific data for reuse in the context of science, subject to a legally guaranteed minimum embargo period enabling data producers to interpret and publish them. This requirement for free data circulation covers firstly publications and secondly data and texts that were not originally scientific but constitute the raw materials of much research, especially research in the human and social sciences.
This requirement to make data available extends to added-value services (massive processing such as Big Data, data mining, relationship with metadata, interoperability) that must also be public and open access to avoid any misappropriation. In the case of creation of services and platforms by publishers and more generally the private sector, this would imply legal guarantees of fair, non-discriminatory pricing.
It also assumes clarification of the authors’ rights to be able to use their scientific productions and publications in relation to publishers and other private actors. The scientists’ intellectual property rights must under no circumstances be transferred to publishers free of charge, so that the free circulation of scientific results can be facilitated.
17The Scientific Board also wishes to acknowledge the work carried out by the COMETS in its Opinion entitled “The ethical issues of scientific data sharing”, and it endorses the recommendations contained in this text.
© OpenEdition Press, 2016