Track A

From ESIWiki

Back to November SAB Workshop

Contents 1 Track A (Newhaven) 2 Questions 3 Malcolm's suggestion 4 Notes 5 Summary

Track A (Newhaven)

Reporter: Jano van Hemert

Democratising e-Science

Easier to use - used by more people - widely understood

AIM: To write the e-Science BlueBook

Questions

• What are e-Science’s grand challenges?
• What are the steps to address those challenges?
• How do we need to change to meet these challenges?
• What is the road map to get e-Science ready for 21st Century challenges?

Malcolm's suggestion

The majority of scientific research is undertaken by large numbers of researchers who back up high-profile leaders. Much is undertaken by altruistic contribution of their time by expert amateur field-workers. Public acceptance depends on public engagement and opportunity to re-use and review scientific arguments.

Would we gain resources, understanding and better e-Science methods by investing more intellectual effort and resources in democratisation? Would we slow the advance of e-Science's front line by using our resources on mopping up operations?

Notes

• e-Science organised in a democratic way or not?
• How is participation organised and policed?
• Happy in the way e-science is being organised?
• Does e-Science work or does it need to be changed?
• Jury out whether e-Science is an elitist venture.
• Benefits and risks with democratising e-Science.
• What kinds of changes are needed?
• Do we think technology in itself can bring the changes? Grid vs Web-2.0 as polarising choices.
• If not just about technology, how to go about it?

Differentiation between Grid and e-Science, suggestion is to not regard Grid for this discussion as it is a technology. Grid clearly an enabling technology, but not the only one, for e-Science and then move the discussion on.

Where do we get funding to move on, and where does the computer scientist fit in the story. Are they one of the scientists or are they the enablers.

We are talking about democratising, should be talking about users, not researchers. Not about the funding question, should be about providing knowledge and data to improve society as a whole. Making science part of everyone's everyday lives. Currently, not possible to publish data and knowledge for others to pick up.

Question of multi layered approace; empowering scientists from domains to create more e-scientists; enabling domain scientists to make use of technology; allowing knowledge transfer.

Issue of intellectual property should be included. Until now assumed academic free flow of information. Is the IPR question a command or a democratising question? Is this the area where the command layer, enforcing the IPR model, can help the democratising process.

MyExperiment is about democratising knowledge in the context of researchers. Main question from researchers, what about attribution in the system? Takes into account copyright, licensing, sharing and ownership.

Science is quite spontaneous, main focus to discover things. Technology just to support them. Problem at the moment is that Grid is trying to cover to many things at once. Web 2.0 is a good model for collaboration. We need definition of e-Science, suggestion is that it is enabling. Also it is evolutionary, as it follows a model of step-wise improvement. e-Science should be routine, not a specific exercise. Start with early adopters, then after it has proved its worth to make it available to everyone.

If that is the vision, what would be on a student's laptop to enable them to become part of democratic e-Science?

e-Science is about access and manipulation of data, not about the computation, which is nothing new. Challenges exist as the amount of data is increasing with an enormous rate. The users should not care about this issue.

Particle physics use case shows that they only wanted to produce something that was good enough to do the job and they did not care about re-use by others.

Question of letting everyone participate may influence the quality of the knowledge and data. Would democratising lead to other ways of doing science, e.g., peer-review processes, engagement of public, etc.? Pragmatic observation, e-Science is not actually addressing these things.

e-Science facilitates the communication and collaboration between computer science and domain science. A list of people with expertise that are willing to interact will be useful. Lubricating that process would be useful.

To do high quality science, not just average, one needs to raise the base. e-Science should therefore facilitate this across a whole community.

Most problems are not seen as challenging for computer scientists. Can we create an environment whether we could solve simple problems on-line. Mailing lists that answer questions are seen as helpful to enable people to use tools. How can we educate the individual domain scientists to distinguish between problems that can be solved trivially and problems that require computer science.

Research community can shown something is technically possible; engineering community should be discussed as they can make it accessible widely. If an ecosystem exists where almost all the bits exist, then it should be possible to bring in those that can facilitate the missing bits.

Dynamic feature of e-Science is that we want to give people the ability rather than provide the solutions every time again.

A healthy ecosystem should encourage people to ask questions, rather than let people work on their own for long lengths of time.

Two directions of discussion. 1. Funding and individual scientists. 2. Mass use and how will students do science in the classroom in the near future.

e-Science started in a democratic way by a selective process. Neither a top-down or bottom-up approach are realistic models. To get work done, e.g., if money is involved, after democracy one needs to select a group of people and say go and decide on this, e.g., come up with a research programme.

Research councils are probably seeing less in e-Science development and want to focus on domain applications.

Discuss the benefits for spreading e-Science around. In some areas it is easy to identify the added value, e.g, the LHC in physics, in other it will be more difficult, such as broad biology.

Suggestion to enable possibilities for domain scientists to connect to computer science.

e-Science and Grid are forever linked because how e-Science came into existence. Hard to convince, for example, scientists in arts and humanities, that e-Science could benefit them. Arts and humanities are not just modifying existing tools from e-Science, they can actually contribute to the development in e-Science.

Can we separate the computation from the rest? E.g., expertise, people, collaboration.

e-Science is really just the tag to say where the funding came from.

Are we in the first cycle of a process similar to what happened in computer aided design, where people now benefit from progress in computer science, as well as keep track of the challenging problems in engineering.

At the start of the e-Science programme, the goal was to make computing resource available on tap to everyone. The problem of not having access to these resources has fixed itself with multi-core systems. The goalpost has changed.

The interaction with the user communities has changed the definition of e-Science itself.

How will you pay for the increasing demands from the various disciplines?

How do you articulate the difficult question: how do you fund the development for that you do not know exist and how can domain scientists ask for it?

We do not know what the benefits will be when e-Science exists out there. We do know collaboration between the disciplines is a key part.

Technology is what allows it, but it is the collaborations that makes it happen.

Can Web 2.0 provide new challenges? They are not perfect and not designed for scientific data; we may be able to improve them using experience from Grid computing.

Grid computing paradigm based on national power grid idea. Has lead to us many empowering battery operated devices, which require much focussed research in batteries. Analogy to Grid is that it is the multi-cores, which will empower us.

Data robustness is an increasing problem as data is increasingly sourced from several locations. Real democracy, demands will be different everywhere. Just have to live with it.

If the success of e-Science is 10% sweat and 90% inspiration, where will the 90% come from? If you can change the world, that will be sufficient inspiration to trigger more investment.

One view that sparked e-Science is that too many computer science departments in the UK had become completely blue sky.

Researchers follow the funding rather than influencing the agenda. However, research councils are open to pressure from science.

What more is the risk of democratising e-science than it is science? Danger of equal access is the quality of data is no longer preserved. Two layers. 1. Data layer 2. Peer-review layer of people that use the data in research. Peer-review may not be scalable to cope with vast increase in data and knowledge.

Finite pot of money available. If spend it on democratising it, we can spend less on "elite e-Science". What is the risk? Depends on who you ask the question, e.g., research councils vs industry. Politics will be in the way when trying to make decisions in a public context.

Mechanisms for democratising. Picking one technology over another, or is that too simplistic? Also, the i dea of the ecosystem exists. Decisions exist of cultural differences and funding issues.

The perfect technology never exists. We make a choice today and change it in five years, while coping with previous decisions. An open discussion should exist about future plans.

Different areas in science have very different requirements. If everyone chooses themselves, they will all choose their own system. This brings a large overhead. Where does the debate take place that forces people to use particular systems (software/hardware/infrastructure). Cost of diversification versus centrally planned. More important to think about the process than come up with the answer directly.

Reasonable to use the "command mode" to preempt this problem and enforce convergence.

Making use of multi-core systems should not be a burden for the end-user and application developer. It is a computer science problem to make use transparent. Chip technology is only one example where barriers exist that prevent people to exploit the technology. A debate on this is visible in Grid vs Web 2.0.

Lack of trust in technology may prove a barrier to adoption.

Security is an important issue; if we do not get it sorted, it may we run the risk of wasting enormous resources on dealing with it later. Lack of trust in the security of the technology is demonstrably an existing barrier - good security is an opportunity as well as a barrier. "good enough" is needed - against a realistic risk assessment.

If this gets out in the wild, natural selection will take over. Also, market mechanisms come into play.

Commercial players and provider which is useful in democratising e-science. What service provider is to be trusted more? Amazon or the UK National Grid Service?

Culture and practice of science. Rewards and points for this kind of research (RAE/Promotion/...) vs "book-based research". This is a problem for adoption. Altruistic work counts for nothing.

Broader problem: no rewards for collaborative research.

A lot of Comp Sci researchers would love to sit and code all day - and could be great at innovating in support of other scientific activity - if they were rewarded for that. (compared to being rewarded for writing papers only)

Interpretation of democratising e-Science. 1. To spread e-Science it around. 2. e-Science as a democratising force. We have to change the culture and make science more democratic to have people gain rewards.

Summary

Questions:

• Is e-Science essentially an elitist venture or has it equally important benefits for the ‘average’ researcher.
• What are the benefits and risks of democratising e-Science?
• How do we proceed, what are the mechanisms and challenges?

Benefits: raising the science base is essential for all science; affording spontaneous collaboration; improving the technology; pushing back the boundaries of participation in science; computing as the sixth utility, this problem has been solved without grids but the emphasis has moved to data and collaboration; lower costs of services and technologies from wider user base; higher quality also.

Risks: implied lack of control, quality issues, etc; will peer review provide the solution? Lack of standards, e.g., for security; opportunity costs – who loses if we channel effort into democratising e-Science?

Barriers: cultural, skills, ease of use; lack of certainty over technology supply, future sustainability; we don’t have the right social architecture to enable wider uptake.

Mechanisms and challenges: ecosystem is an attractive metaphor for how grid and web 2.0 might interact but how would it work; is there ever a ‘right’ technology to pick? if we chose the wrong technologies there are communities who will not get engaged, e.g., security; tech supply; we need to engineer mechanisms to reward participation, encourage embedding, e.g., education and training; one possible social architecture would involve encouraging, e.g., a market to complement (or even replace) centralised, top-down provision of services and technologies. What are the appropriate business and funding models to encourage this?