Realities of Software Technology Payoffs

1
Realities of Software Technology Payoffs

• Kevin M. Burns

2
Personal Selection Process

• Read a variety of abstracts and picked 3 topics
  of interest
• Datawarehousing
• Three-Facet View of Information Systems
• Software Development Process

3
Why Software Development?

• A software product is often behind schedule, over
  budget, non-conforming to requirements and of
  poor quality.
• Fragile and insecure software continues to be a
  major threat to a society increasingly reliant
  on complex software systems.

4
Original Paper

• The Realities of Software Technology Payoffs
  by Robert Glass
• Essence
• What does research tell us about the value of new
  technologies rather than, how we can improve
  research in order to learn more about the value
  of new technologies.
• Approach Evaluative research

5
Evaluative Research

• In evaluation research a researcher formally and
  objectively appraises the outcome of a project,
  program, strategy, and so on. The researcher also
  carefully examines the chief factors that are
  positively or negatively influencing the
  performance of the activity. The primary
  objective is to gather data to facilitate making
  a decision regarding the future of the activity.
  Evaluation researchers employ a broad range of
  both qualitative and quantitative research
  methodologies. Such research is most commonly
  carried out in business, educational and
  nonprofit organizations, as well as governmental
  agencies.

6
Software Development

• Is it
• Simple and well understood lending itself to
  dramatic improvement Basili and Fenton
• Complex and hard with little to no breakthrough
  claims - DeMarco

7
Background

• 1980s Emphasis on Software Productivity
  Improvements
• 1990s Emphasis on Software Quality
• Promise of orders of magnititude improvement

8
Improvements

• Structured Techniques
• 4th Generation Languages
• CASE
• Formal Methods
• Cleanroom Methodology
• Process Models
• Object-Orientation

9
Findings
10
Are there any conclusions that can be made?
11
Evaluative Research To Date

• Most improvements to the software process offer
  modest advantages over traditional alternatives.
• May not offer breakthroughs but they do show
  things are on the right track for a long time.

12
My Additional research

• Rather than choose an area of interest like OO I
  chose the technology that holds the most promise
  based on the article.
• Process Models
• Controlling and improving the process to develop
  software

13
Process Models

• One example
• Software Engineering Institute (SEI) at Carnegie
  Mellon
• Software - Capability Maturity Model (SW-CMM)
• Another example
• ISO 2001

14
Cited in Original Paper
15
Recap CMM

KPAs
16
Recap CMM
17
Recap CMM
Each KPA has a set of goals, capabilities, key
practices, measurements and verification
practices. Goals and key practices can be used
to assess the impact of a KPA. Total of 52 goals
and 149 key practices.
18
Additional Paper

• Clark, Bradford K., The Effects of Software
  Process Maturity on Software Development Effort
  Source USC Center for Software Engineering PhD
  Dissertations, University of Southern California,
  August 1997.
• Dissertation

19
Clark Paper

• Assumption
• Effort is the primary determinant of software
  development cost and schedule. The technical
  challenge in this research is determining how
  much change in effort is due solely to changing
  Process Maturity when this change generally
  occurs concurrently with changes to other factors
  that also influence software development.

20
Clark Paper

• 3 available approaches used by models to measure
  effort
• Analogy not applicable to Process changes
• Theory not applicable due to aggregation of
  input variables
• Statistics best for Process Maturity if inputs
  are callibated and correlated
• Multiple Regression Analysis isolate a single
  response variable from predictor variables.
• Log-Log model nonlinear into linear and then
  apply linear regression analysis.
• The most influential factor is the size of the
  software product. Others are complexity,
  experience and supporting tools.

21
Clark Paper

• The problem
• SW-CMM is a specification of
• What should be in the software processes
• Does not answer how or when
• Roadmap and/or set of evaluation criteria
• Benefits usually described in terms of
  productivity, a controversial measurement

22
Clark Paper

• Productivity (Boehm)
• Ratio of outputs produced by the process to the
  inputs consumed by the process
• What constitutes inputs and outputs?
• Instead of productivity
• Clark measures Process Maturitys effect on
  effort, which is a fundamental component of
  productivity

23
Clark Paper
Predicator Variables of Effort
24
Clark Paper

• Clarks Hypothesis
• Increasing the level of Software Process Maturity
  decreases the amount of software development
  effort
• KPAs primary contribution to effort is rework
  (Hertsleb, 1994) ie. changing requirements,
  poor planning
• Instead of case studies, as had been
  traditionally used,
• Collect and Analyze data to quantify factors that
  effect effort, including Process Maturity

25
Clark Paper

• Research Model

26
Clark Paper

• Uses 6 mathematical models that support the
  following conclusions
• For all 112 projects, Software Process Maturity
  was a significant factor (95 CI).
• After normalizing for other effects, a
  1-increment change in the rating of Process
  Maturity resulted in 15-20 reduction if effort.
• Modeling approach used in this analysis can be
  used in other areas of Software Engineering.

27
Summary

• Ideas for areas of additional research
• Is effort the primary cost and schedule driver?
• Combinations of Improvement Techniques
• More emphasis on quality studies as opposed to
  productivity studies
• Qualitative Research and Mathematical Models
• Question Answers