The focus of our discussion in the Global Systems Science “narratives” workshop has been on the narrative as a lens onto a model, for consumption by decision-maker and citizen. I’d like to make a connection with a related discussion in scholarly communication – where the narrative is instead for the scholar, but some of the issues are pertinent, particularly to do with the narrative as a social object and digital object.
There is much discussion in the scholarly communications (and future of research communication) community just now about “Beyond the PDF”. One approach is to ask how the academic paper (a mechanism about 350 years old) evolves with modern digital practice. Another – and this is my provocation – is to ask what will be the shared digital artefact that scholars will be exchanging in the future?
There is already evidence of new practice and new objects – for example, aggregations of data and procedural knowledge which may be executable. These research objects are compound digital objects but also social objects around which discourse occurs and social networks form, and they are produced and consumed by humans and machines: they typically contain narratives.
What does this mean for us? I suggest four points:
- We should consider if narratives are also to be consumed and produced by machine, and if this is achieved through text processing or bundling with machine-processable forms. Even if not machine-generated, the lifecycle of narratives might surely be machine-assisted;
- The social life of narratives is an interesting thing to instrument and analyse; e.g. their provenance, usage, evolution. If nothing else this helps us use narratives more effectively, but also it enables analysis of collaboration in the complex sociotechnical ecosystem that is Global Systems Science;
- As we think of narratives throughout their lifecycle we can think also of their inter-relationships and associations with the other digital artefacts of Global Systems Science, such as the models, the experiments, the dataflows, …;
- Research Objects themselves may be of interest, as they are a mechanism for sharing methods, for reproducible science, for automation to handle scale and assistive systems to enable human creativity – all things we need for Global Systems Science.
A closing thought re (3). One criticism of papers is that they enforce exchange of “human sized chunks of knowledge” and are only targetted at specific audiences, so might actually act to constrain our science. A model that is bundled with multiple narratives might serve better, behaving as a boundary object which can be exchanged between communities – with a common core and multiple interpretations for different users.
I shall mention some of this in my talk Knowledge Infrastructure for Global Systems Science in the Information Society, Models and Narratives session on Saturday morning. For more on the Future of Research Communication check out FORCE11, and there is an emerging literature on research objects.
— Dave
Professor David De Roure
Director, Oxford e-Research Centre
UK National Strategic Director for Digital Social Research
University of Oxford
Additional points from my talk:
1. What are the social objects of GSS? Based on our recent discussions, these include models and narratives.
2. How do we achieve automation that is assistive and scales? A risk of Taylorisation is that we burn in scientific assumptions, also it may be more important than ever to label all results as “provisional”. Computational Research Objects are a step in this direction and suggest a research agenda in their own right.
3. Social Machines for Systems Science. A reflexive observation – we are building social machines to conduct GSS, and also we study them as part of GSS.
An interesting tension played out in subseqent discussions. The Web 2 approach perhaps encourages “publish then filter” (or as we say in software, “release early release often”), whereas there was significant concern that we have to be very careful what is put out to the public (caricatured as “filter then publish”). In fact we have seen this in myExperiment, where social statisticians are nervous about sharing models (lest they be applied without being properly understood) while bioinformations seem comfortable with indiscriminate sharing of workflows.
There are good examples of the value of opening scientific results for interpretation by the crowd, and indeed this is a principle of the government open data initiatives. As Patrick pointed out, this is about different ways of building trust – by authority or by the crowd.
— Dave
PS I feel has been an outstanding event. Elsewhere we see tensions between disciplines, but here we have the disciplines engaged in a sophisticated common endeavour with deep acknowledgement of the multidisciplinary challenges, critical reflection and appreciation of the nuances (and without the obstruction of disciplinary ego). This is all essential to the success of this community.
[I’m catching up on GSS matters after a period of teaching.]
I really like this post – for any global system there will have to be many interacting parts (models) and these artifacts (research objects) really need to be not only human-readable but also machine-readable. For modelling in the large the interfaces between (sub-)models is very important and we need good formalisms, domain specific languages and tools to do “computer aided modelling”.
/Patrik
http://www.chalmers.se/cse/EN/people/jansson-patrik
http://www.gsdp.eu/about/the-consortium/prof-patrik-jansson/