Designing for Palpability in e-Research

1
Designing for Palpability in e-Research

• Alexander Voss
• alex.voss_at_ncess.ac.uk
• National Centre for e-Social Science and
  e-Science Institute
• With Rob Procter, Mark Hartswood, Mark
  Rouncefield and Roger Slack

2
The Grid Vision of Seamlessness

• a hardware and software infrastructure that
  provides dependable, consistent, pervasive, and
  inexpensive access to high-end computational
  capabilities.
• (Ian Foster and Carl Kesselman)
• Like the power grid, the Grid makes services
  available through common interfaces without the
  user having to worry about the details of how
  these services are provided.

3
Grid Resources as Commodies?

• There are various reasons why the notion of
  access to Grid resources as undifferentiated
  commodities is problematic
• Resources come in any number of types and are
  provided by multiple, independent resource
  providers
• There is a lack of common (technical and
  organisational) standards for the management of
  access to these heterogeneous resources
• Software licensing issues often hinder the use of
  compute resources on a large scale
• Users of datasets in particular need to
  understand what their quality attributes are, how
  they may be used and how data may or may not
  change over time

4
Scientific Workflows

• Scientific workflow systems such as Taverna or
  Kepler allow scientists to define orchestrations
  of services providing access to resources on the
  Grid.
• Workflows are an important instrument not only
  for automation but also for elaborating and
  documenting the research process.
• Direct manipulation or form based input make this
  functionality available to researchers who need
  not acquire programming skills.
• Complex social relations are forming around the
  development and (re-) use of workflows in groups
  and communities.

5
Workflows are Potentially Fragile Artefacts
6
Scientific Workflows

• Workflows are complex composite processes that
  are long-running and subject to possible failure.
• There is, therefore, a need for people to be able
  to reason about their properties, to investigate
• State and runtime behaviour
• Interaction with services
• Correctness
• Failure modes and recovery
• Provenance of data produced
• This kind of work is required prior to, during
  and after the execution of a workflow.

7
Adding Collaboration to the Picture

• Of course, if we are not merely talking about
  technical assemblages, things get more
  complicated.
• Awareness What are people up to?
• How can we provide resources for people to
  account for the activities of others as well as
  for whats going on in technical systems?
• Provenance as well as practical questions
• Adding issues of privacy and confidentiality
• We may want to know what data is used by whom and
  for what purpose

8
Palpable Computing

• ubiquitous/ambient computing complemented with
  palpable computing
• invisibility - visibility
• construction - de-construction
• scalability - understandability
• heterogeneity - coherence
• change - stability
• automation - user control and deference
• Büscher et al. Bottom-up, top-down? Connecting
  software architecture design with use. Available
  at http//www.ist-palcom.org

9
Conclusions

• Palpability is a potentially powerful concept for
  the design of e-Research technologies and
  infrastructures
• There is a tension between the vision of seamless
  operation of the Grid, various technical and
  organisation issues and the need to reason about,
  manage and document the provenance of research
  outputs
• e-Research poses significant challenges (a
  non-exhaustive list)
• Lack of control over services
• Different, potentially conflicting policies
• Potentially vast scale
• Long-running, costly operations
• Need to predict, steer and document computational
  behaviour
• Work in wider collaborative contexts spanning
  various boundaries
• What additional challenges would pervasive
  e-Research pose?