Grammars of Collaboration: Designing for e-Science Mark Hartswood, Roger Slack, Kate Ho, Marina Jirotka, Rob Proctor, Jenny Ure, Alex Voss.

Grammars of Collaboration:Designing for e-Science

Mark Hartswood, Roger Slack, Kate Ho, Marina Jirotka, Rob Proctor, Jenny Ure, Alex Voss

Vision and Reality

• One role of visions is to provide a future orientation for research and practice; they can sometimes, however, be blind to the sorts of practical problems on the ground which impact on its realisation

• Quantitative and qualitative changes• Scientific work and scientific communication• Situated and virtual• Local and Global• Social and Technical

• Everyday interactions on the ground that shape and are shaped by these new ‘virtual organisations’ and in many cases hinder the realisation of the vision

• Examples from a number of Grid based projects

Examples from eHealth and others

– eDiamond – GS: Scottish Family Health Study– MRC NeuroGrid– NTRAC: National Translational Cancer Research

Network (Edinburgh Centre)

So-called ‘joined-up’ systems envisage services being delivered through virtual organisational structures (VOs)

Flexible VOs formed around networks within, and across, multiple service units and administrative domains

The Vision of the Virtual Organisation

• Across disciplines• Across organisations• Across CoPs• Across complex

distributed human and technical networks

Translational Medicine

Patient Care

Bench Science

Drug Development Epidemiology

Clinical Trials

‘From bench science to clinical practice’

A

Survival time post diagnosis (years)

876543210

Cum

ula

tive S

urv

ival

1.0

.9

.8

.7

.6

.5

.4

.3

.2

.10.0

Carriers

Non-carriers

Edinburgh NTRAC CentreIntegrating clinical, molecular and trial data

Vision of Benefits

• Shorter start-up periods for studies, cost-effectiveness and earlier realisation of outcomes

• Feeding the virtuous circle of translational research• Getting benefits of e-Science projects realised in practice• Technologies that are ‘in working order’:

– in line with NHS infrastructure– in line with research infrastructure– usable in clinical and research contexts

• Platform for eHealth innovations• Direct benefits for patients through trials and feedback of

research results

Gap between vision and reality

Relating ‘bleeding edge’ research to established, routine, accepted practice requires (among other things)negotiation of obligations, expectations, reciprocities associated with sharing of data and resources in local communities

Data Integration : the NeuroGrid Vision-the social life of information

• Integration of data collected for very different purposes

• Reliability of data collected across multiple sites – or even across the same research lab

• Myth of shared protocols!

Subject groups, Trial purposes, Trial dataLongitudinal studies over several yearsDifferent scanners, protocols, clinical/cognitive testsDifferent data formats

Varying methods and regions of interestAlgorithms such as Freesurfer, SPM, auto-Gyrification IndexVarying clinical diagnoses and demographics

Differences across CoPsDisciplines -Psychiatry, Psychology, Computer Science, Neuroscience, Physics, Radiology,

NursingAims-funding Strategies – competition vs collaboration-Criteria – cost, time, usability

Implications for Grid-based VOs

‘One might say that Grid technologies represent a shift from data and resource sharing in collaboration as a craft or cottage industry, to something that can be routinely engineered and expected to behave in a well mannered way’

Implications for making collaborative work visible

in virtual organisations

Local GrammarsThe articulation of local community structures is an intrinsic part of the

social process in natural communities

• Shared understanding • Shared aims and criteria • Shared and visible mechanisms for carrying out,

Providing additional technical infrastructure can make performance worse if the social, technical and socio-technical articulation of the complex is not in alignment.

Increasingly, system design reflects the need to generate a similar process for larger ensembles that do not have the shared spaces in which to do so.

Supporting project collaboration

• Developing embryonic community infrastructure as basis for co-creating a socio-technical one.

• Shared spaces• Shared frames of reference

Nokia Arrabianranta

Socio-technical & Socio-political Grammars

• Vision of Grid science dependent on socio-political, legal and contractual infrastructures not yet in place

(NH Records)

• Resulting tensions affect realisation of the translational science vision e.g. tensions between ethical consent and research access to patient records in eHealth

e-Science & scientific process

• Gives rise to new ‘virtual organisations’ (Foster & Kesselman, 2004)• More heterogeneous• More interdisciplinary• More potential for alignment and misalignment (examples)• Opportunities for rethinking the nature of scientific work• Recurring problem: solution scenarios

Aligning the whole and the parts: visualisation

Interest in the different ways in which VOs can shape or be shaped by the grammar of collaborative processes in local contexts

• Role of mapping these (often invisible) local processes to inform design

• Role of designers in making the processes in the VO more visible for the users

Visualising systems: allowing users to ‘see’ the implications of action in the system

Visualising data architecture for users

Building systems around the cognitive process.

• WebSOMs• Shneiderman• Bush• Pask• Hitchens

Visualising local processes for designers

eDiamond Involved ethnographic

studies of collaborative process ‘in the wild’ with implications for a virtual infrastructure to extend that

The collaborative process in the wild

• Computer-aided Detection (CAD)– Use image analysis software to detect potential abnormalities– Draw these to the reader’s attention using a ‘prompt’– Designed to prevent readers from overlooking a possible abnormality– Has a number of potential roles:

• Making screening more sensitive• Supporting single reading • Supporting less experienced reader

Decision-aids in mammography

• The idea is for prompting systems to act as attention cues • Look at the images and reach own conclusion before looking at the

prompts• However, we saw evidence of prompts being used as decision-aids:

“I’m not really that worried about it. [At all?]. But as CAD’s marked it now, it’s a case of – do I really take more notice of it? … I’ll mark it. I’m going to mark it down - as possibly being something.” (transcript from video)

VOs heighten the need for synergy & alignment to common ends

•One size fits all

•Global and Local Requirements

•Federated Local Requirements

Local and Global collaboration

• Software designed to standardise safety compliance procedures globally, was actually increasing risk in some local operating sites

Aligning heterogeneous and distributed communities of interest

Collaboration can add value

Or cost and risk

• Challenger

• Iraq procurement system was deemed a success - technically

Tension

Interviewer:

You’ve mentioned the problem of requirements ‘creep’ late in the design. Can you think of anything that might have helped avoid this?

Technical Manager:

‘A cluster bomb perhaps?’

Grammars of consent, liability,reward

• Grid protocols for acceptable use of resource• Ethical consent for use, reuse, repurposing• New or varied conversations became possible

for which these rights, permissions and potential benefits or penalties have not been negotiated and for which a process is required

• The e-Science bundle of new paradigms, technologies and concepts has challenged the accepted order that is seen to govern how collaborations conventionally unfold in less distributed contexts.

• Making the collaborative process more visible to designers and users is part of realising the Grid vision

Barriers to Grid Vision

Collaboration in designing systems was about criteria and reward within

particular communities as much as knowledge transfer

Many of the problems were recurrent scenarios found in other Grid projects, and in other distributed socio-technical systems

Visions of eScience: the ‘third way’

• Buetow (2005) suggests that the cyber-infrastructure provided by and for e-science can reconfigure our perceptions of what doing scientific research in distributed settings might be

• Laurillard• VLE ebusiness experience

Users face problems understanding

• Provenance of data• Reliability of data• Security of data• Implications of action – who sees the data etc• Dependability of service• Shape of the organisation

Transformational Technologies?• Emergent work practices and requirements may only

become evident as users attempt to apply the system to their work

• Requirements capture and design are currently separated off from the deployment of the system.

• Through ‘learning by doing’ and ‘learning by interacting’, users are able to experiment, share and appropriate the innovations of others, mobilising their collective resources to evolve systems, to continue ‘design-in-use

Nokia and Arrabianranta

Visualising systems: allowing users to ‘see’ the implications of action in the system

Future Work

• Policies that govern the VO are codified and embedded in the collaborating systems, and interactions between the organisation are audited.

• This provides an opportunity to visualise the VO to end users

• aim is to explore how existing e-Science infrastructures could be used to meet these usability requirements

Recurring Collaborative Strategies in other Systems

• Map existing process• co create a new one

Building Technology Around Social Processes

• Local Scenario• SSM• CATWOE• Amazon• Limewire• eBAY - brokerage

Using the Architecture of Social Networks

• Brokerage• Closure• Burt• Sense-making• Social Capital

Pre-requisites for Collaboration

• Shared spaces• Shared frames & terms

of reference• Shared aims

• The ‘file’ ‘programme’ analogy

TeTechnical Social

Open

Closed