© colin potts b2-1 small-group approaches to requirements determination colin potts georgia tech
TRANSCRIPT
© Colin Potts B2-1
Small-group approaches to requirements determination
Colin Potts
Georgia Tech
© Colin Potts B2-2
The role of group work in requirements
Multiple stakeholders & perspectives» Need for consensus-building procedures
Unstructured & ill-understood problems» Need for methods for recording & keeping
track of unstructured information Planned change
» Need for facilitation & participation All of these are small-group issues
© Colin Potts B2-3
Overview: small group techniques
Brainstorming» Proposal & evaluation of new ideas through
facilitated meeting practices Issue tracking
» Keeping track of loose ends through systematic note-taking
Joint Application Design (JAD)» Reaching consensus on new system
requirements through workshop discussion
© Colin Potts B2-4
Brainstorming for desired features
Goal: produce many ideas quickly» Application to reqts: identifying features or
alternative automation strategies Golden rules
» Produce lots of ideas, minimize discussion Benefits
» Rapid creation of team approach & shared appreciation of ramifications
» Most applicable to product design or situations where the customer has very vague requirements
© Colin Potts B2-5
Brainstorming procedure
Select group carefully Phase I: Divergence (brainstorming proper)
» Generate as many ideas as possible in fixed period» Record all ideas on shared writing space» Build on or combine previously posted ideas, but
don’t discuss or criticize
Phase II: Synthesis» Cluster similar ideas and combine» Rank and select ideas (e.g. using weighted voting)
© Colin Potts B2-6
Class exercise: Brainstorming
Take the requirements document & imagine you are starting the project that wrote it
Using brainstorming, identify 50 desirable product features in 10 minutes
Discuss how to collate/combine the features Vote on priorities of the remainder
» You can vote for 20 features» Features with 10% vote are retained
Discuss the quality of the resulting product
© Colin Potts B2-7
Brainstorming: Where to find out more
Brainstorming, and team-based idea generation methods like it are summarized in a number of books:» Jones: Design Methods» Scholtes: The Team Handbook» Gause & Weinberg: Exploring
Requirements
© Colin Potts B2-8
Issue tracking
Design and planning answer questions» “Is” questions
– “What do you do in the library?”– “What exactly is a patron ‘in good standing’”?
» “Ought” questions– “How do you want this information to be communicated?”– “Is this feature more important to you than that?”
Questions can be addressed to all stakeholders» Customer, actors, owner (in SSM terms)» May be answered by them or on their behalf
© Colin Potts B2-9
Issues and actions in meeting minutes
Keeping track of actions is standard
Can also keep track of open issues» Some of these are
actions, if one person has responsibility to resolve
– But others are more open-ended & just need to be remembered
It was decided to base the meeting scheduler on a shared calendar.
Action: Frank to sketch 1st-cut schemaIssue: How much of one’sschedule can a co-workerknow?
For example...
© Colin Potts B2-10
IBIS: Issue-based information systems
Structured method for keeping track of issues, positions and arguments» Originated in architecture & urban planning
Issue» An open question about which there are likely to be
opposing points of view
Position» A response to an issue by a stakeholder
Argument» A reason put forward by a stakeholder for choosing or
rejecting a position
© Colin Potts B2-11
Example IBIS graph
Issue: How much of one’s schedule can a co-worker know?
Position: Only whether one is free
Position: All details of appointments
Argument: Important to retain privacy
Argument: We have no secrets here
Argument: Information may be useful in scheduling
+
+
-
© Colin Potts B2-12
QOC: Design space analysis
Similar to IBIS, but uses explicit criteria Compare:
Question: How much of one’s schedule can a co-worker know?
Option: Only whether one is free
Option: All details of appointments
Criterion: Privacy
Criterion: Compatibility with corporate culture
Criteria: Spin-off benefits
+
+
-
+
-
0
© Colin Potts B2-13
Experiences with issue tracking
NCR has used IBIS for restaurant IS design
Design Space Analysis has been used to record design rationale for interactive systems at Xerox» See Moran & Carroll book for more
information In practice, the ideas are watered down
© Colin Potts B2-14
1.1 The system shall blah, blah...1.2 If the co-worker is blah, blah, thesystem shall inform the user ...1.2.1 Blah, blah, blah......
The Inquiry Cycle (1)
Insight: Questions are always prompted by something» Attach questions to the document fragments that
gave rise to them
Question: How much of one’s schedule can a co-worker know?
This questionannotates aspecific reqt.
© Colin Potts B2-15
The Inquiry Cycle (2)
Answering the question should lead to a revision of the artifact
1.1 The system shall blah, blah...1.2 If the co-worker is blah, blah, thesystem shall inform the user only that the co-worker is busy...1.2.1 Blah, blah, blah......
Question: How much of one’s schedule can a co-worker know?
Decision: Only whether one is free
This decision recordrepresents the rationalefor the new reqt.
© Colin Potts B2-16
Team exercise: Issue tracking
Take a requirements document and review it in an issue-based way.» Attach several issues to the document at specific
places» Propose more than one position in response to
each issue» Either list arguments for or against the positions,
or list criteria that the positions contribute to or detract from
As a class, discuss what insights arise
© Colin Potts B2-17
Issue tracking: how to find out more
IBIS and Design Space Analysis:» A good anthology of articles
– Moran & Carroll: Design Rationale
Inquiry Cycle» Introductory article
– Potts et al: Inquiry-Based Requirements Analysis
© Colin Potts B2-18
Joint application design (JAD)
Structured approach to setting scope for system using facilitated workshop» 3-10 day JAD session» “Session leader” facilitates» IBM developed JAD in commercial sector
Three phases(1) Preparation
(2) The session (3-10 days)
(3) Feedback
© Colin Potts B2-19
Assembling a JAD team
Roles» Executive sponsor» JAD leader» Scribe» Support person» Full-time user
participants» Full-time MIS participants» Users on call» Observers
Guidelines» Full-time attendance is
mandatory– minimize distractions
» 8-15 participants– Everyone is equal
» 2-3 users for every MIS participant
» Participants must have representative knowledge & authority
» Mgt. commitment is essential
© Colin Potts B2-20
Preparation for JAD
Project definition» Identify open issues & interview mgt.» Select team & schedule session
Research» Familiarize with system & document workflow» Gather preliminary specifications
Session preparation» Preparing working document» Pre-session meeting
© Colin Potts B2-21
The JAD session
Typical agenda:(1) Review assumptions
(2) Review current workflow & determine new one
(3) Define data elements, screens & reports
(4) Record & resolve open issues
Making decisions» Failing consensus, use weighted criteria (cf. QOC)
Open issues» Assign responsibility & deadline to resolve
© Colin Potts B2-22
After the JAD session
Making the working document authoritative» Incorporate all decisions & issue for approval» Session members review for clarity & accuracy
– but not for content
Approval of the document» Important to have authorizing managers (user &
MIS) sign off
Post-JAD changes» Working document should be living document
© Colin Potts B2-23
Team exercise: JAD planning
Imagine you are planning a JAD for the example system. In teams of 2-3:» Identify who should attend the JAD
– name names if possible
» Estimate duration of JAD session» Identify sources & types of information during
research phase» Record some open issues
Discuss the activity as a class
© Colin Potts B2-24
JAD: how to find out more
The best book on JAD is» Wood, J. & Silver, D. Joint Application
Design, Wiley, 1989» It describes the JAD process in detail,
giving advice on what to do» It also gives advice on more generic
meeting facilitation challenges
© Colin Potts B2-25
Cooperative requirements capture (CRC)
A JAD-like process» Smaller group (6-9)» Same research-then-improve structure» Two 2-day meetings
– Understand business problem– Propose improvements
» Uses many sheets/forms to record current/future issues
More information:» Macaulay’s book
© Colin Potts B2-26
Participatory design (PD)
In PD, user representatives are members of design team» Unlike JAD, participation continues throughout
project» Origins in Scandinavia
Emphasis on representation of system using mock-ups and low-fidelity prototypes» Less emphasis than JAD on documentation» PD often uses JAD-like “future workshops”
© Colin Potts B2-27
Future Workshops in PD
Preparation» Designers become users’ apprentices
Future workshop» Fantasy phase
– Brainstorming & weighted voting about future HAS
– Metaphors (e.g. library as a warehouse)
– Write ‘utopian outline’
» Implementation phase– Prepare common strategy based on utopian outline
Build prototype
© Colin Potts B2-28
Envisionment through scenarios
Scenarios are descriptions of actual or imagined sequences of events
Scenarios Specifications
Concrete Abstract
Representative orsalient examples
Exhaustivecoverage ofpossibilities
Specific General
To illustrate withcompelling examples
To definethoroughly andunambiguously
© Colin Potts B2-29
Dimensions of scenarios
Concreteness» Can the scenario branch or not?
Detail» e.g. “the patron borrows a book” or “the patron presents a
book and card, the librarian swipes the card, and...”
Instantiation» e.g. “Diane borrows ‘War and Peace’”
Representation» Narratives, interaction diagrams, rich pictures, etc.
Mood» Does it describe current or desired system?
© Colin Potts B2-30
Tabular representation of scenarios
Role Action
Initiator Call mtg
Scheduler Send msgs to all invitees
Typical invitee Read msg
Typical invitee Consult personal schedule
Typical invitee Respond with desired times
Scheduler Collate desired times
Scheduler Select best time
Scheduler Notify invitees and initiator of mtg time
time
© Colin Potts B2-31
Scenarios as stories
Q: How do we select the “right” scenarios to explore?» A: They tell interesting stories about the system
– e.g. normal cases, interesting exceptions & combinations
Like folktales/myths, named scenarios persist» E.g. later on the team might ask “how would this
work with Joe’s meeting?”
© Colin Potts B2-32
Team exercise: Scenario invention & analysis
For the example system» As a class:
– Brainstorm some scenario names– Pick one for further analysis
» In teams of 2-3:– Describe scenario at high level in table– Re-do it in more detail
» As a class:– Discuss any insights about the reqts.
© Colin Potts B2-33
Scenarios: how to find out more
Book» Carroll: Scenario-Based Design
Articles» Special section in ACM SIGCHI» Potts et al Inquiry Cycle article
© Colin Potts B2-34
The importance of facilitation
In all these methods, effective facilitation of the group is essential» Precise structure of the activities is less
important than ability to work together & converge
» Activities best done in meetings, not through email or memos
Facilitation requires practice & experience
© Colin Potts B2-35
Small group approaches: conclusions
Goal:» Have key stakeholders converge on vision of
future system & agree on essential details
Differences:» Degree & duration of user involvement» Structure of activities vs. structure of process
Key requirements:» Time and commitment of mgt. & participants» Good facilitation & preparation