Thus, the total effort required is 75 man-months, together with the use of 47 hours of machine time. The total cost of the exercise would therefore be approximately £25,000 spread over a period of a year. This expenditure is not entirely over and above existing expenditure because, in effect, Levels 1 and 2 of the proposed procedure are already carried out by the present procedure. Thus, the new expenditure required is about £13,000, i.e. a doubling of current expenditure.
10. Consequences of approval
For the procedure to be effective, there must be some financial consequences of the software not obtaining approval. A possible approach is to attach 10 per cent of the purchase price or rental to each level of approval. This would allow up to 40 per cent of the price to be retained under late delivery conditions. It would also be necessary to require at least a minimum set of software being available before any money was paid. A possible approach would be for the contract to list the required items of software and against each item give the minimum level of approval required for acceptance of the installation. It will also be necessary to have associated with the contract the specifications of the required software. These would detail the facilities and performance of the software. However, this will require a change of attitude on the part of computer manufacturers who are often very reluctant to provide any detailed performance information until after the software in question is working. This attitude requires the customer to purchase his software almost as an ‘act of faith’ in the supplier and is surely not something that can be allowed to continue.
11. Conclusions and discussion
This paper has attempted to present a possible method by which large organisations, such as the United Kingdom government, might test the software supplied by computer manufacturers to its installations. Its main aim may be described as an attempt to ensure that new computer installations can rapidly take on the work for which they were purchased by ensuring that installations have the best available information on which to plan their work. It is a procedure aimed at elucidating the facts about software; what works and what does not and how well does it work? The cost of doing this sort of exercise is not trivial but with the number of computer systems involved, the cost per installation is quite small and if the procedure can speed up the take-on of work by allowing the confident use of proved software, rather than the hesitant use of the unknown, it will prove worthwhile. Hopefully, such a procedure would encourage the computer manufacturers to improve their own quality assurance procedures, perhaps even to the point where use of testing of software becomes unnecessary.