The closing plenary was presented by Dan Atkins, W.K Kellogg Professor for Community Information and Associate Vice-President for Research Cyberinfrastructure, University of Michigan. His topic was the status of the e-science movement in both the UK and the US. In introducing this theme, he explained that there needs to be a matrices of relationships, approaches and sets of goals, so he would be weaving between three different perspectives: the funders, researchers and universities.
Atkins observed that e-scientists are starting to address computational discovering and computational thinking for all at various service levels, which includes modelling, simulation, prediction, exacting knowledge from data of all kinds. He felt that the UK has been ahead in this area. However, he was keen to pick up on a comment from one of the earlier speakers, who observed that “no good crisis should go unused”. Infrastructure is generally taken for granted except in times of crisis, and currently many US institutions are having to look very hard at the amount of money they are spending on cyberinfrastructure. Some are just dictating cuts, whilst others are not only looking for efficiencies but are also trying to look at the mission effectiveness of ICT, consolidation of demand and sourcing of services. He observed that this means that ICT enabled collaborations that are now possible will only work if the end user is engaged and people are increasingly looking for this outcome. He emphasised that the ultimate goal is no longer just high performance computing, but also high performance collaboration.
To focus in on this, Atkins quoted the UK definition of e-science by Sir John Taylor, which highlights collaboration and information utility. He noted that this contrasts with the evolution of the e-science programme in the US.
Atkins continued by discussing the experience of chairing an NSF Blue Ribbon advisory panel, which recommended that the definition should be approached in a more agnostic way with respect to technology and should not be so tightly coupled to the notion of the grid. He assured us that the e-science movement in the UK had been very influential in the US and was one of the factors which led to the call for the Blue Ribbon review that he chaired. Their resulting report led the NSF to create a new office and a reflective report that embraced the recommendations that they made.
He went on to discuss the resonance between the push of the technological capabilities and the pull of the scientific need by observing that e-science or science enabled by cyberinfrastructure and ICT is increasingly essential for meeting the grand challenges of 21st century science. This is because there is an inherent complexity, multi-scale and multi-science nature of todays frontiers, increased scale and value of data and demand for semantic federation, demand for active curation and long-term preservation of access. In the US, may be less in the UK, these investments are critical to education, so these same investments could be made to create opportunities for students and teachers to engage in more high quality, authentic, passion building science.
One of the results of the report was the reflective report from the NSF called Cyberinfrastructures for the 21st Century. This was, for the time, quite revolutionary as it had more in it than high performance computing. The report called for a multi-component approach that involves high performance computing, but also recognised the need for cyberinfrastructure as an object of study and a mechanism for enabling learning.
Atkins went on to strongly recommend the book The Fourth Paradigm: Data-Intensive Scientific Discovery. This points out that we have had science for thousands of years based on empirical observation, we recently added a theoretical branch, then a computational branch then latterly a data exploration branch, which is the fourth paradigm.
Atkins observed that increasingly people are understanding that the e-science movement is real and growing and critical. It is now being recognised as relevant to the humanities and arts, and is being coupled with e-learning – not just putting textbooks up, but taking advantage of the whole participatory element. The term e-development has also arisen. Atkins emphasised that the whole future of the research university will be influenced by this movement, and they should pay attention to the e-science movement because it is effectively a pilot of the techniques which will be relevant to the whole of a research university.
He then took us through some of the findings and recommendations of the UK e-science review, which included:
- The feeling that the UK is already in a world-leading positions
- That the investments so far have already empowered significant contributions in the UK and beyond
- That the UK is at a cross roads and must decide whether to create the necessary combination of financial, organisational and policy commitments to capitalise on their prior investments
- That e-science is an organic, emerging process, requiring ongoing, co-ordinated investments
The key recommendation was that the UK should continue to nurture a robust infrastructure.
Atkins took us through the many strengths of the current programme and what it had already achieved, but also highlighted the weaknesses, which included the too rapid and early reduction of core funding and high level leadership. This left to a sense of abandonment. He also noted that there is a lack of applicable models for sustaining and maintaining software/infrastructure operations – often due to funding restrictions. He used the Moores revised technology adoption life cycle diagram to identify that there is a chasm between the momentum of the early adopters and the early majority, left by the withdrawal of funding support.
Atkins moved on to discuss what the universities doing about cyberinfrastructure, citing the recommendations outlined in a report entitled: Research Cyberinfrastructure Strategy for the CIC: Advice to the Provosts from the Chief Operating Officers. Based on this, he observed that universities should all currently be investing in:
- preparing for federated identity management
- maintaining state of the art communication networks
- providing institutional stewardship of research data-intensive
- consolidating computing resources, whilst maintaining diverse architectures
To illustrate this, he discussed the activities of the University of Michigan, which include developing an IT visioning, planning and governance model. This features mission stewards and domain stewards who take responsibility for understanding the needs of the institution. Faculty representation and student representation. He observed that there are often fundamental issues of governance and who decides about cyberinfrastructure, and many institutions do not have the kind of model that clearly identifies roles and jurisdictions.
Atkins then introduced the University of Michigan’s CIRRUS project, which focusses on exploring the use of pod optimised performance data centres, which have much better energy efficiencies. The current situation in research computing involves facilities over the whole university, and researchers who have their own machines. This project is trying to migrate more to their FLUX machine and cloud capabilities. Part of the CIRRUS project is to understand the options and which options fit best from a research and a finance position.
Finally, Akins discussed the issues facing the NSF and the main challenges that they are currently working on. Key to this was an understanding for the need to approach all cyberinfrastructure as part of an ecosystem rather than a set of several entities, which linked in neatly with the main themes of the conference as a whole and helped set the agenda for the future with of the NSF in the e-science arena.
Further details and references from Professor Atkins’ talk can be found in his slides.