KG: As a researcher, my need is to perform quantum calculations that usually take many days or weeks to complete, and require memory capacity and consistent stability from HPC, rather than speed. That said, I’m completely agnostic about the underlying system that meets these needs. Most HPC work is actually done by post-grads, however unlike physics post-grad who like to engage with the HPC system, chemists are just looking for an efficient, stable service. The impression is that you need a high level of expertise to use HPC in your researcher, but that’s not true in every discipline – our post-grads just need HPC as a stable service.
KG: Collaboration is no longer a choice for scientists. As a researcher in chemistry, you’re not going to be effective unless you collaborate, simply because people have had to become so specialised. To make impactful scientific insights, you need to bring complementary skills together. Links between disciplines are required in order to deal with multi-disciplinary problems.
KG: Collaboration for many scientists is based on self-confidence in their own ability. Less collaborative people are usually less confident. That said, most younger staff are very keen to collaborate, as they know it makes their papers better and allows them to compete on an international scale. People need to think differently than they have in the past. Cultural change is required if researchers are to begin considering the bigger picture implications and applications of what they’re doing and who they could be collaborating with.
KG: I think the competitive funding model used by government tends to discourage collaboration. There is too much focus on ensuring competition between what are very small players in a global context. NZ institutions should be collaborating to compete at an international level, not divided and contending for funding at home. If we funded sport in NZ the same way we fund science, the All Blacks would be a pretty poor rugby team.
KG: The challenge for scientists is to maintain productive collaboration that is connected to the world. We need stronger connections from theorists through experimental research to practical application and all the way out to industry. In this respect, the government has been really clear – they want to see a translation of the National Science Challenges into a future where scientists take a more collegial approach to addressing research questions that will make a real difference to society. They want people to live healthy, productive and fulfilling lives, in a safe environment – and they expect scientists to collaborate to contribute to this goal,
KG: Sometimes our organisations have very different systems that might limit collective goals (for example, Auckland University and Otago University take completely different approaches to managing access to and paying for large capex instruments). Another tension between institutions is how we manage the issue of organisational data, especially research data, when it’s not clear what it is, where it is, who owns it, who manages it, and how specific it is to individual organisations and departments. Shared eResearch infrastructure forces scientists to work together and offers real opportunities to overcome some of these differences; however we always need to remember and respect that our various institutions are different.
KG: I don’t really believe there’s any true data security – if you put a safe in your house to protect your valuable, all you’re really doing is slowing down the thieves long enough to discover them. For this reason, I don’t give much credence to ideas of data sovereignty – where you store your data has almost no relationship to how safe it is, or whether it’s going to be there in the long term. If longevity and reliability are the goals with data, then the large off-shore providers are probably always going to do that better than any home grown solution. NZ suppliers can’t really offer guarantees of security or longevity, as they don’t have the scale to support this. Precedents make a big difference. More examples of successful sharing of infrastructure will lead to more inclination to plan capex in a collaborative manner.
KG: Using off-shore providers doesn’t mean we give up control and become a price taker. The Pharmac model of dealing with very large multi-national providers in the health industry appears to have worked well for NZ (even if the multi-nationals don’t much care for it). Maybe we can use this model to negotiate with commercial off-shore providers for a deal for all of NZ’s research sector.
KG: In shared infrastructure, there are some aspects that really are distance sensitive and aren’t as effective unless you’re located close the instrument – tissue sample research for example. Certain equipment such as MRIs or spectroscopes need to be close to where the samples / bodies are – so that limits the ability to share these investments. This is not the case with HPC, as the nuts and bolts can be anywhere. What researchers need to be local and available are the people and expertise that can help us make best use of the shared HPC resource.