A new RAD-based approach to commercial information systems development: the dynamic system development method
Alan Howard Senior Lecturer, School of Information Management, Leeds Metropolitan University, Leeds, UK
The dynamic system develop- ment method (DSDM) is an holistic approach to system development in a rapid appli- cation development (RAD) environment. It provides an overall framework, based on wide experience, for particu- lar types of commercial information system develop- ment projects. The method is currently undergoing trials in a number of user organiza- tions. The formation of the consortium and the success- ful launch of the method shows how user organizations and IT professionals can take a proactive stance in the advancement of system development practice.
According to Weinberg’s Law, “If builders built buildings the way system developers build computer-based business systems, the first woodpecker that came along would destroy civilisation”! A 1996 survey by accountancy firm Coopers & Lybrand in eight countries including the UK concluded that two thirds of IT projects in the previous two years were late, cost more than the budget or failed to meet requirements. Clearly some- thing is amiss. Information systems profes- sionals within the DSDM consortium believe they know what it is.
Traditional system development is usually conducted in a sequential fashion. Require- ments are explored and analysed before a system to meet them is designed. If accepted, the design is implemented and tested. Eventu- ally, when users, clients and developers are satisfied, the system is made operational. Requirements are often “frozen” at the design stage so that exact specifications for system components can be documented.
projects. If this is extrapolated throughout the rest of industry and commerce the scale of the problem is clear.
The dynamic system development method is proposed as a way to alleviate and over- come problems with traditional development. DSDM is a non-proprietary RAD method which has been developed by capturing the experiences of a large number of user and vendor organizations. It provides a frame- work for building and maintaining computer- based systems which are subject to tight development timescales. The method not only addresses the developer view of RAD but that of other parties in effective system develop- ment: project sponsors, users, project man- agers and suppliers.
The consortium was formed in January 1994. Version 1 of the method was released in February 1995. Version 2 was published in November 1995. Trials of the method have been taking place in several member organi- zations. Early feedback has been promising and lessons learned will be incorporated into future versions.
Industrial Management & Data Systems 97/5  175–177
© MCB University Press [ISSN 0263-5577]
Static, sequential development is subject to and causes many problems. Business and IT strategies change as managers come and go and new technology changes our perception of solutions. Requirements determination is plagued by communication problems. Users find it difficult to explain their requirements in terms that developers can understand. Technical staff find it difficult to understand business requirements. In any case require- ments are subject to continuous change as businesses, customers and markets develop. System design tends to exclude users. Con- struction is labour intensive and costly. As implementation deadlines approach, testing is frequently rushed or abbreviated. When the system is installed users are often disap- pointed and feel little commitment to the systems produced. No sooner is the system installed than corrective and enhancement maintenance is required to make the system usable and acceptable. Often IT systems do not even reach the final stages and are aban- doned.
Another recent report indicated that three UK life insurance companies have written off £75m in the past five years on failed IT
Principles of DSDM
A fundamental assumption of DSDM is that nothing is built perfectly first time, but that 80 per cent of the solution can be produced in 20 per cent of the time it would take to pro- duce the total solution. The premiss is based on the nineteenth century economist, Pareto’s Law, which asserts that it is the final 20 per cent of any problem solution that takes the majority of the problem-solving effort. It is the exception processing routines and extra features that increase development time disproportionately. Basic functionality can usually be accommodated relatively eas- ily in IT projects. By focusing on rapid deliv- ery of the basic functions, workable systems can be produced quickly, thus reducing the cost and time of projects.
Along with this assumption, dynamic sys- tem development is based on a number of underlying principles which dictate how a project is undertaken. Version 1 identified 13 principles. In version 2 these have been reduced to nine as follows:
[ 175 ]