A National Research Data Service for the UK?

I attended the A National Research Data Service for the UK? meeting at the Royal Society in London last week and my live-blogged notes are available for those who want more detail.  Chris Rusbridge also blogged the day on the Digital Curation Blog - session 1, session 2, session 3 and session 4.  FWIW, I think that Chris's posts are more comprehensive and better than my live-blogged notes.

The day was both interesting and somewhat disappointing...

Interesting primarily because of the obvious political tension in the room (which I characterised on Twitter as a potential bun-fight between librarians and the rest but which in fact is probably better summed up as a lack of shared agreement around centralist (discipline-based) solutions vs. institutional solutions).

Disappointing because the day struck me more as a way of presenting a done-deal than as a real opportunity for debate.

The other thing that I found annoying was the constant parroting of the view that "researchers want to share their data openly" as though this is an obvious position.  The uncomfortable fact is that even the UKRDS report's own figures suggest that less than half (43%) of those surveyed "expressed the need to access other researchers' data" - my assumption therefore is that the proportion currently willing to share their data openly will be much smaller.

Don't take this as a vote against open access, something that I'm very much in favour of.  But, as we've found with eprint archives, a top-down "thou shalt deposit because it is good for you" approach doesn't cut it with researchers - it doesn't result in cultural change.  Much better to look for, and actively support, those areas where open sharing of data occurs naturally within a community or discipline, thus demonstrating its value to others.

That said, a much more fundamental problem facing the provision of collaborative services to the research community is that funding happens nationally but research happens globally (or at least across geographic/funding boundaries) - institutions are largely irrelevant whichever way you look at it [except possibly as an agent of long term preservation - added 6 March 2009].  Resolving that tension seems paramount to me though I have no suggestions as to how it can be done.  It does strike me however that shared discipline-based services come closer to the realities of the research world than do institutional services.

Comments

[Lorcan Dempsey]

You raise the question of scalar emphasis which is a growing issue across a wide range of our activities; whether we are thinking of geographic or organizational scales. It is at the heart of the Information Environment discussion for example which was a topic here a few days ago.

At what scale should certain things be done?

The Institution is clearly the wrong place for quite a bit of activity which currently goes on there. You, for example, have argued that institution-scale repositories represent misplaced attention/energy.

Discovery poses an interesting question. Where discovery has become web-scale what is the role of the institutional or national apparatus?

For what things is a national-scale solution sensible? What makes sense at the institutional? Where does neither make sense.

These are primarily oganizational/business questions which is why, incidentally, I am slightly surprised to see the IE discussion start from technical contexts.

[David Groenewegen]

I'm not convinced that institution-scale repositories are a waste of energy, as long as they can be connected to the discipline "space". That is, you store the data at an institution (which has a vested interest in it, and a permanence that other options lack) but you connect that to a discipline based front end (like TARDIS - http://www.tardies.edu.au) for working with those in your field.

I agree that trying to force open data sharing is pointless, and that we need to work with those who want to. We also need to distinguish very carefully between types of sharing - sharing with a select group of colleagues in "secrecy" (ie within a closed virtual environment), sharing by making the metadata about a dataset widely available but only allowing access to those who contact you and you establish a relationship with, and publishing the dataset online openly are all good things, and need to be supported. If the latter becomes the sole definition of "data sharing" not much will occur.

[Rachel Heery]

I agree that activity around research data would benefit from lessons learnt from our experience so far with repositories for research papers - not least that user behaviour needs to be central to any innovation. As you say building on existing data sharing seems a good way forward.

Whilst recognising that in some ways research can be characterised as 'global', in other ways it is 'individual' and 'institutional'. I most definitely think institutions have an important role in research as well as in any sharing and curation of outputs. The institution is a funder of research in so far as it supports the individual researcher, and the institution will benefit from a researcher's success. So the institution has a significant interest in the way research outputs are shared and curated, and could have an important role in any cultural change. Whilst that does not signify that any 'solution' needs to be institutionally based , to my mind it does mean that institutions are significant potential agents for change. Successful innovation at the institutional level might transition to activity at a different 'scale' - a bit like cataloguing did.

[bentoth]

I came across the report of this meeting by chance, and the thing that struck me about it is the complete absence of a clinical trials/Health R&D/NHS/Department of Health/NIH perspective. Perhaps this was by design - I hope so really because if not it was a missed opportunity.

A few things about Health R&D - it's massive (NHS expenditure on R&D is in excess of £1bn annually); important (Clinical Trials) and has already demonstrated real world benefits of sharing and combining data (Early Breast Cancer Trialists Collaborative). There are also specific units in the UK with a remit to collect and curate clinical trials data (Clinical Trials Units)

The experience of clinical trials also shows how hard it is to standardise data - it would be a mistake to imagine that 20 years of standardisation efforts going back to Ida Sim's work and more recently BRIDG has produced any significant breakthroughs.

There is a new round of clinical trials data standardisation and curation starting but it seems to be happening in a parallel universe.