The period between the 1970's and 1980's was a time of great advancement in computer
hardware technology which took an industry still in it's infancy, to a level of much
sophistication and which ultimately revelutionised the information storage and processing
needs of every other industry and that of the entire world. However, it was also during
this period when the shortcomings of implementing such technology became apparent. A
significant number of development projects failed which resulted with disastrous
consequences, not only of an economic nature, but social aswell. Seemingly, although
hardware technolgy was readily available and ever improving, what was inhibiting the
industry was in the methods of implementing large systems. Consequently, all kinds of
limited approaches materialized that avoided the costs and risks inherent in big-systems
developments.
Times have changed, and with it our understanding and experience as how best to develop
large systems. Today's large systems yield greater benefits for less cost than those of
previous decades. Large systems provide better, more timely information, the ability to
integrate and correlate internal and external information, the ability to integrate and
facilitate streamlined business processes. Unfortunately, not every system that
information workers develop are well implemented; this means that the computer system
which was originally intended to make a company more efficient, productive and
cost-effective, is in the end doing the exact opposite - namely, wasting time, money and
valuable manpower.
So even with all the lessons learned from the 70's and 80's, our vastly superior
methodologies and knowledge of the 90's is still proving to be fallible, as suggested in
the following examples.
System Development Failures
In Britain, 1993, an incident occurred which forced the London Ambulance Service to
abandon its emergency system after it performed disastrously on delivery, causing delays
in answering calls. An independent inquiry ordered by British government agencies found
that the ambulance service had accepted a suspiciously low bid from a small and
inexperienced supplier. The inquiry report, released in February 1993, determined that
the system was far too small to cope with the data load. For an emergency service, the
system error would not only cause the loss of money, but more essentially, fail to
dispatch ambulances correctly and promptly upon the arising of critical situations.
Thus, the implications of such a failure are apparently obvious, both socially and
economically. Since the failures, the ambulance service has reverted to a paper-based
system that will remain in place for the foreseeable future.
Another failure was the collapse of the Taurus trading system of the London Stock
Exchange. Taurus would have replaced the shuffling of six sorts of paper among three
places over two weeks - which is how transactions in shares are settled in London-with a
computerized system able to settle trades in three days. The five-year Taurus development
effort, which sources estimated cost hundreds of millions of dollars, was termed a
disaster, and the project was abandoned in March 1993. Exchange officials have
acknowledged that the failure put the future of the Exchange in danger.
Why did they fail?
What went wrong with these systems? The real failure in the case of the London Stock
Exchange was managerial, both at the exchange and among member firms. The exchange's
bosses gave the project managers too much rope, allowing them to fiddle with
specifications and bring in too many outside consultants and computer firms. Its new
board, having heavy-weight and diverse membership, proved too remote from the project.
Member firms that spent years griping about Taurus's cost and delays did not communicate
their doubts concerning the project. The Bank of England, a strong Taurus supporter,
failed to ask enough questions, despite having had to rescue the exchange's earlier
attempt to computerize settlement of the gilts market. According to Meredith , an expert
in project management issues, many system development catastrophes begin with the
selection of a low bidder to do a project, even though most procurement rules state that
cost should be only one of several criteria of designation. The software failure occurs
because the companies involved did not do a risk assessment prior to starting a project.
In addition, many companies do not study the problems experienced in earlier software
development projects, so they cannot apply that data when implementing new projects.
Another source of problems is the failure to measure the quality of output during the
development process. Information workers as yet have not fully understood the
relationship that exists between information and development. It is shown that
information should be viewed as one of the essential know-how resources. The value and
necessity of information for development is argued. An attempt is made to classify the
various areas where information is needed for development, as well as the information
systems and infrastructures available or required to provide for the different needs.
There are a number of reasons why information has not yet played a significant role in
development. One reason is that planners, developers and governments do not yet
acknowledge the role of information as a basic resource. Another is that the quality of
existing information services is such that they cannot yet make an effective contribution
to information provision for development.
Avoiding development failure
Companies blame their unfinished system projects on such factors as poor technology,
excessive budgets, and lack of employee interest. Yet, all these factors can be easily
avoided. All that is needed to develop and implement successful systems is a strong
corporate commitment and a basic formula which has proven effective time after time. By
following the guidelines below, any system workers can install and implement a
successful, efficient system quickly and with minimal disruption to the workplace.
Understand your workplace-every company must fully understand its existing environment in
order to successfully change it.
Define a vision for the future- This objective view will help the company develop a clear
vision of the future.
Share the vision- In order for the system to be successful, all those who are involved in
its development must fully buy into the process and end-product. This will also help
further define specific goals and expectations.
Organize a steering committee-This committee, which must be headed by the executive who
is most affected by the success or failure of the project, has to be committed and
involved throughout all stages.
Develop a plan-The project plan should represent the path to the vision and finely detail
the major stages of the project, while still allowing room for refinement along the way.
Select a Team of users- A sampling of company employees is important to help create, and
then test, the system. In the Laboratory systems failure case . That means both the
vendor and laboratory should identify what users know and what they need to know to get
the best out of the LIS. They must also develop a formal training plan before selecting a
system.
Create a prototype-Before investing major dollars into building the system, consider
investing in the development of a prototype or mock system which physically represents
the end product. This is similar in concept to an architect's model, which allows one to
actually touch and feel the end product before it is created.
Have the users actually develop the system- It is the end-users who will directly benefit
from the system, so why not let them have a hand in developing it? In the DME is DBA case
, the fault that the Open Software Foundation(OSF) make it's Distributed Management
Environment system fail is the OSF tried to go from theory to perfect product without the
real-would trial and error that is so critical to technology development.
Build the solution-With a model in place, building the solution is relatively easy for
the programmer. Users continue to play an important role at this stage ensuring smooth
communication and accurate user requirement.
Implement the system-Testing the system, training and learning new procedures can now
begin. Because the majority of time up until now has been spend planning and organizing,
implementation should be smooth and natural, and most importantly quick.
The Role of SAA and ACS in the Assurance of Quality
The Standards Association of Australia was established in 1922 as the Australian
Commonwealth Engineering Standards Association. Their original focus was on engineering,
subsequently it expanded to include manufacturing standards, product specifications and
quality assurance and consumer-related standards. The role the SAA play is in quality
certification. According to SAA, a standard is a published document which sets out the
minimum requirements necessary to ensure that a material, product, or method will do the
job it is intended to do. For systems development, both the Standards Association of
Australia and Australian Computer Society give the guides and standards to develop a
system and to control the quality of a system and to prevent failure from occurring. They
also make the standard of the system developed connectable world wide.
When software development projects fail, they usually fail in a big way. For large
development projects, the cost is typically astronomical, both in terms of dollars spent
and human resources consumed, some with even further reaching implications effecting
adversely the whole of a society. Too often, mistakes made in developing one project are
perpetuated in subsequent ones. As with the error which occurred in the London Stock
Exchange system, what they should have done was find out how the system allowed the error
to happen and fix it, then learn from it for making better developments for future
information systems.
Bibliography:
1. Fail-safe Advice, Software Magazine, George Black, 3/93
2. All fall down, The Economist, Anonymous, 20/3/93
3. DME is DBA(Dead Before Arrival), Data Communications, Robin Layland, 2/94
4. There's No Excuse for Failure, Canadian Manager, Selim EI Raheb, 9/92
5. Laboratory Systems failure: The enemy may be us, Computers in
Healthcare, Stanley J. Geyer, M.D., 9/93
6. Australian Standard Software quality management system, Standards
Australia
|