Pages of History
Pages of History
Philip T. Ransact*
University of Oxford
Only now seen in museums and special collections, the academic paper was a primary means of intellectual discourse amongst scholarly communities of old. Invented four centuries ago, the physical document format largely fell into disuse after the turn of the millennium. While digital versions continued in certain disciplines for another three decades, the format was recognised to be slowing the pace of innovation and was eventually superseded. Could this lengthy and expensive transition have happened sooner?
Inventions have long been documented in writing (the first British patent was issued in 1618) and proving a claim of first discovery is one incentive for recording scientific work. Another is that science can best move forward through transparent and open exchange of ideas accompanied by experimental evidence. This latter principle, quite revolutionary at its time, underpinned the publication of the “Philosophical Transactions of the Royal Society” journal by the Royal Society of London in 1665.
Fast forward to the 1990s when the World Wide Web was designed to allow physicists to share scientific news and information. It enjoyed phenomenal uptake and with it societal transformation. However we may observe that in science – the very domain of the Web’s ambition – the paper remained much the same, just in digital form. It was a remarkably resilient scientific and social object for over three centuries.
The End of The Paper
The turn of the new millennium also saw a computational turn in research and the rise of “digital scholarship”. Where research once involved a few people and a few computers, now it involved “big” and real time data and huge social networks. The world’s grand challenges demanded large, distributed, multidisciplinary teams and long-running experiments. Human triage and analysis of growing dataflows were increasingly assisted by automation through software and “cyberinfrastructure”. The days of the paper, like its pages, were numbered.
The demise of the paper around 2030 can be attributed to several factors:
- It was no longer possible to include the evidence in the paper.
- It was no longer possible to reconstruct a scientific experiment based on a paper alone.
- Writing for increasingly specialist audiences restricted essential multidisciplinary re-use.
- Research records needed to be readable by computer to support automation and curation.
- Single authorship gave way to casts of thousands of collaborators and citizen scientists, leading to failure of the authorship and incentive model.
- Quality control models scaled poorly with the increasing volume and “open access” movement, obscuring innovation.
- Alternative reporting was found necessary for compliance with increasingly stringent scientific and industrial regulations.
- Frustrated by inefficiencies in scholarly communication that stifled progress, research funders demanded change.
A Period of Confusion
The scholarly publishing system was an innovation bottleneck and clearly needed innovation. And here is the catch-22: the techniques for innovation were prone to that very bottleneck. The problems were recognised, many scholarly meetings were held and much was discussed, but progress was inevitably slow and incremental. Scholarship suffered a “Self-Defeating Innovation Bottleneck” .
A very positive step occurred in the 2010s with the recognition of the importance of research data, but it failed to embrace the many new artefacts of digital research. Within the research ecosystem new objects were being exchanged: software, workflows, scripts, policies. Probably the practices of open source software best mirrored the changing practice of research, so it was doubly due of attention.
Meanwhile considerable confusion arose in the crisis of reproducibility. This tenet of the scientific method, based on independent reconstruction of experiments, suffered when the sharing of digital artefacts seriously interfered with the notion of independence: sharing was at once beneficial and self-defeating.
An important move “beyond the document” came with the introduction of various kinds of Research Objects , which described the bundling of artefacts needed to interpret or reconstruct research and were designed also for machine consumption and maintenance. The boundaries, sharing, mutability and encapsulation of research objects were fluid, but this facilitated co-evolution and important insights were gained through use.
Clarity and Co-Constitution
Papers survived as long as they did because they had something right: they contained human-sized chunks of knowledge, by humans and for humans. Trying to bundle the infrastructure into the paper was bound to be problematic. The last two decades have seen a welcome return to narratives so that we can communicate all aspects of research between scientist, citizen and policymaker alike.
No longer is it “one size fits all”, and nor do we conflate human narrative, intellectual content and executability. Perhaps the most important outcome of co-evolution has been the emergence of new agreed ways of expressing the design of our science systems – the configurations of tools and resources that constitute the interacting information circuits we use and re-use on an everyday basis.
We still have human-sized chunks of knowledge, represented at the right level for human consumption and reasoning and deeply linked. What was once called “the literature” is an increasingly machine-processable research record. The “semantic silos of selfish science” are becoming a thing of the past, and credit comes from accurately tracking the provenance of ideas, not just counting packets of communication.
Today we more closely achieve the desired symbiosis whereby computers do what they are good at and free humans to most effectively do their piece. We accomplish research at scale in many dimensions, by understanding that research requires a sense-making network of social machines and social objects – with innovation and without reinvention.
Papers were fantastic but very much of their time. It took an astonishing 20 years to end the “tyranny of the paper”, resulting in much inefficient research investment. Could we have seen this sooner? Perhaps engaging other disciplines would have helped:
- Engineering understood how to design, model and specify systems – a circuit diagram from a century ago is reproducible today, thanks to the right levels of abstraction and some of the obligations of working in the physical, not digital, world.
- Social Science and Humanities (and especially the field of Digital Humanities, predating the millennium) understood the role of the human, the philosophy of science, the nature of interpretation and creativity, and the risks of Taylorisation.
Papers were, however, an important paradigm of scholarship for at least 350 years and will continue to provide important ‘reading material’ for historians.
The author wishes to thank colleagues from the “History of Research Communication and e‑Scholarship” community (HORCE61).
- D. De Roure (2013). Pages of History. Beyond the PDF 2. Amsterdam, Netherlands. March 2013.
- S. Bechhofer, D. De Roure, M. Gamble, C. Goble and I. Buchan (2010). Research Objects: Towards Exchange and Reuse of Digital Knowledge. Nature Precedings, doi:10.1038/npre.2010.4626
*Illustrative pseudonym. This format, showing a single author of a paper with affiliation to a physical institution, was typical of academic papers. Only the named authors received reward for ‘citations’ and further funding for their research.