A Cognitive Perspective On Boundary- Spanning IS Design Dr. Susan Gasson Assistant Professor

advertisement
A Cognitive Perspective On BoundarySpanning IS Design
Dr. Susan Gasson
Assistant Professor
College of IS & T
Drexel University
An Investigation of Boundary-Spanning IS Design
 The nature of organizational IS design and the role of
boundary-spanning knowledge management in high-level
design.
 Traditional (decompositional) process model vs. convergence
model of design process.
 Three views of design as social cognition.
 Research study: participant observation and ethnographic
data collection of boundary-spanning design process in
organizational context over period of 18 months.
 [If time] SSM as a tool for surfacing implicit understanding.
 Research findings: the nature of a boundary-spanning design
process.
Boundary-Spanning, Enterprise-Level IS Design
 Organizational IS design is viewed here as a high-level,
conceptualization process:
 The giving of form to an organizational IS
 Involves the co-design of business and IT systems
 Distinct from the low-level “design” stage of SDLC.
 Involves knowledge sharing and negotiation of consensus
across multiple
IS
Bid
knowledge domains or
manager Financial
process
organizational boundaries.
accounting
manager
Extent of shared
understanding
manager
Product
Marketing
Operations engineering
manager
finance manager
manager
Traditional Model Of Design Process
Consensus on
organizational
problem and goals
for change
Gap analysis: process of design
Agreed form of IS
solution
 Individual, rational model of problem-solving (Alexander, 1964).
 Assumes consensual, objectively-defined set of initial goals
(Simon, 1973).
 Empirical studies reveal emergent strategies:
 “Opportunism” of expert software designers (Guindon, 1990).
 “Improvisation” in designing IT-related organizational change
(Orlikowski, 1996).
Convergence Model of Design
Framing of
organizational
change problem
Framing of
organizational
change problem
Framing of
organizational
change problem
Framing of
organizational
change problem
Framing of
organizational
change problem
Framing of design process goals (gap analysis)
Framing of
target system
solution
Framing of
target system
solution
Framing of
target system
solution
Framing of
target system
solution
Problem Framing
of organizational
Framing of
change
target problem
system
&
goals
solution
Framing of
target system
solution
No longer goal-driven, but continual evolution of “gap analysis”
between how we understand (frame) the problem and how we
understand (frame) the solution.
Research Question
 Does the convergence model offer a convincing
alternative to the decompositional model of design
and, if so, how does the convergence of problemand solution-space take place in boundary-spanning
group design?
Boundary-Spanning IS Design Process
• Convergence model still deals with individual processes –
does this explain collaborative process?
• People are members of multiple social worlds, through
their membership of different work and disciplinary
groups (Strauss, 1983; Vickers, 1974).
• Organizational "problems" not consensual but emerge
through interactions between the various social worlds to
which decision-makers belong (Suchman, 1998; Weick,
1998).
 Therefore, we need to examine processes of social cognition,
to understand collaborative design process.
3 Views Of Social Cognition
 Socially-situated cognition:
 Situated action (Suchman, 1987; 1998): shared work-spaces are
produced through social and contextual interaction; continually
redefined.
 Cognitive “frame” (Goffman, 1974) – “structures of expectation” guide
how people predict and interpret context (Tannen, 1993).
 Socially-shared cognition:
 Shared frames: cognitive "shortcuts" provide shared interpretations of
organization without the need for complex explanations (Boland and
Tenkasi, 1995; Fiol, 1994).
 Congruence between “technological frames” (Orlikowski & Gash,
1994).
 Distributed cognition (Hutchins, 1995):
 Understanding is not so much shared between, as "stretched over"
members of a cooperative group (Star, 1989).
 Coordination achieved through “heedful interrelating” (Weick &
Roberts, 1993).
Research Questions
 Does the convergence model offer a convincing alternative to
the decompositional model of design and, if so, how does the
convergence of problem- and solution-space take place in
boundary-spanning group design?
 IS design as socially situated cognition:
 How do individuals' design frames interact, to form a group
"framing" of an information system?
 IS design as socially-shared cognition:
 Does a design group develop a shared design-frame over time?
If so, what aspects of the design are shared?
 IS design as distributed cognition:
 How does a boundary-spanning design group manage and
mediate distributed design understanding?
Boundary-Spanning Collaboration in
The Co-Design of Business and IT Systems
Participant observation & ethnographic data collection of group of 7
managers involved in design of IS to support the process of responding to
customer invitations to bid for new business:
Commercial
division
Contractual
policy
Process
Improvement
Manager
Bid
response
Bid
response
Financial
strategy
Finance
division
IS
Manager
Core design
team
Bid
response
Product
development
Technical
division
Marketing
Customer division
interface &
intelligence
Bid
response
Bid
response
4
Product/
customer
strategy
Production
capacity
planning
Cost/effort
estimation
Operations
division
Framing in IS Literature
 Concept comes from cognitive psychology
 Orlikowski & Gash (1994) used concept of “Technological
Frames” to represent different understandings of the role of
technology in work.
 Davidson (1996, 2002) extended concept to understand how IT
system stakeholders understood what IT is required and the
role that this would play.
 Problem of granularity: this is behavioral and not cognitive
research:
 Frame congruence (Orlikowski & Gash, 1994) – frames are
similar in content and structure (qualitatively coded).
 Problem of explicit vs. implicit knowledge about IS:
 Framing involves implicit knowledge, so employed Soft
Systems Methodology (Checkland, 1981) to elicit.
Data Analysis Methods
1. Qualitative coding of “levels of problem
decomposition”
 Analyze goal-orientation and decompositional focus
of group process
 Qualitative coding of “contributions” to design
meetings at 4 points, distributed throughout design
project.
2. Qualitative coding of in-depth interviews, based on
SSM (Checkland, 1981), to derive design “framing”
concepts employed by individuals.
3. Discourse analysis of design meeting transcripts to
understand how distributed understanding was
managed.
Research Question
 Does the convergence model offer a convincing
alternative to the decompositional model of design
and, if so, how does the convergence of problemand solution-space take place in boundary-spanning
group design?
Design Was Not Decompositional
Meeting
A
B
C
D
Episode of
design
1
3
5
6
Purpose of meeting
overall system purpose & functions
detailed design of stage 1
detailed design of stages 2-6
implementation of stages 2-6
Intended level of
decomposition1
4-5
3
3
1-2
Average level of
decomposition
3.28
3.05
2.75
2.82
High-level design goals
5
4
3
2
1
Detailed implementation mechanisms
Predicted
Actual
Simulated Single Meeting
Research Questions
 Does the convergence model offer a convincing alternative to
the decompositional model of design and, if so, how does the
convergence of problem- and solution-space take place in
boundary-spanning group design?
 Sub-questions, from a “framing” perspective:
 How do individuals' design frames interact, to form a group
"framing" of an information system?
 Does a design group develop a shared design-frame over time?
If so, what aspects of the design are shared?
 How does a boundary-spanning design group manage and
mediate distributed cognition?
Data Analysis Methods
2. Discourse analysis, to derive design “framing”
concepts employed by individuals



Data collected (in interactive interviews) at 3 points:
 beginning,  middle (approx.), and  end of
design project
Interview questions focused on 3 aspects of design,
reflecting the 3 elements of the convergence model:
 Problem-framing
 Solution-framing (target system goals and form)
 Process-tasks required to get from problem to solution
(gap analysis)
Employed Soft Systems Methodology techniques
(Checkland, 1981) in interviews, to surface implicit
frames/understandings.
SSM Goal
Surfacing:
6 Definitions
of “Achieve
Higher
Quality In Bid
Process”
Engineering Manager
Takes an inordinate amount
of time to respond to Bids
which was not allowed for in
resource plans
Senior management
acknowledge that servicing
Bids should affect business
as normal
Bid Process Manager
Bid-process centred on Bid
Manager’s role
Bid process handled by
business as a whole
Senior Finance Manager
High degree of informality in
Bid process
Process Improvement (Quality) Manager
Business position at start of
business process redesign
project, struggling to cope
with volume of orders
Project Management Accountant
Narrow focus on getting
business
IS Manager
Individual authoring tools
used to generate Bid
response sections, so no
consistency
More managed Bid process
Business has changed:
volume is no longer the
issue; quality and
presentation of response is
the issue.
Wider focus on getting
business (new customers
and/or new people and/or
new technology specified
Consistent ‘look and feel’
achieved for all sections of
Bid responses
Example SSM Root Definition For Process-Task
Project boundaries are to
limited to fully achieve
objectives of process
Root Definition: A system owned
by the Managing Director where
the Team leader identifies which
processes need analyzing and
obtain resources for this for the
benefit of the people who operate
the Bid response process and
customers.
This is necessary because the Bid
response process is bound up in
other business processes.
It is constrained by the
subsequent impact on wider
processes which interface with
any process that interfaces with
the Bid process.
Those processes which
interface with Bidding
process are improved
Identify appropriate
people to analyse these
processes
Identify processes
which bound Bid
response process
Allocate processes
Identify someone to own
process improvements
Identify mutual
problems with Bid
response process
Report on progress and
gain management buy-in
for required changes
Management:
monitor progress in problem resolution
Monitor progress in
tackling mutual
problems
Implement required
process changes
How do individuals' design frames interact, to form a group
"framing" of an information system?
 Organizational problem frames
 noticeable convergence of individuals’ framing-perspectives.
 Target system goals
 converged to some extent towards the middle of the project
 appeared to converge superficially towards the end
 use of common metaphors, e.g. "electronic document library"
 but differed widely in meanings attached to metaphors
 overall, little convergence at the level of individual
understanding
 Design tasks required (gap analysis)
 diverged widely at all stages.
Extent of shared
understanding
IS
Bid
manager Financial
process
accounting
manager
manager
Product
Marketing
Operations engineering
manager
finance manager
manager
Does a design group develop a shared design-frame over
time? If so, what aspects of the design are shared?
 The use of shared metaphors in defining system goals or
aspects of a solution did not indicate a shared understanding of
what those goals/solutions entailed.
 What was shared was an understanding of how the problem on
which the group was focusing was structured
 e.g. the use of the phrase “the big-arrow, little-arrow concept”
indicated a shared understanding that the team’s problem was to
find a way of defining the Bid process so that it was aligned with,
but separate and parallel to the product lifecycle process.
 But problem definitions were discarded when they caused
perceived dissonance with individual frames.
 Replaced by more sophisticated problem-structure that
embodied some elements of previous structure, but also
replaced other elements.
Does a design group develop a shared design-frame over
time? If so, what aspects of the design are shared?
Initial problem definition
Modified problem definition 1
Modified problem definition2
Modified problem definition 3
.
.
.
SHARED PROBLEM
DEFINITION
Actual path of design
VAGUE TARGET
SYSTEM GOALS
Perceived path of design
At each change in direction:
Emerging information about
organizational processes
Existing problem definition
Changing
Project
Scope
Revised, organizational
change goals
Revised design
problem definition
Discarded, partial
problems & goals
• Very different from the traditional model,
design was driven by shared problem-frames, not shared goals.
How does a boundary-spanning design group manage and
mediate distributed design understanding?
 Through understanding problems in common, the
group was able to develop sufficient levels of trust
that they could delegate responsibility for parts of the
solution definition, when this was too complex for
one person to understand in detail:
 "I know that Peter wants to fix the same things that I
want to fix, so I'll trust him to sort out his end of the
system [personnel training]".
How Do Groups Manage Distributed Knowledge In Design?
 Shared attitudes and beliefs towards the design (why are we
doing this and how do we want to change the company? –
built through developing shared problem-structures) guide
shared interpretations of the organizational environment.
 This permits groups to negotiate distributed understanding of
design tasks and how to perform them.
 Negotiation is facilitated by the use of boundary objects (e.g.
design models), that capture and communicate a joint
knowledge of the design that is greater than the knowledge of
any individual.
 Knowledge of who knows what allows group to distribute
work effectively among themselves.
 Appears to be established through individual “specialization” in
specific application-domain areas, during the design process.
Research Question
 Does the convergence model offer a convincing
alternative to the decompositional model of design
and, if so, how does the convergence of problemand solution-space take place in boundary-spanning
group design?
Questions?
Download