If producing usable systems is a cost then an argument needs to be produced to show that the benefits of a usable system outweigh that cost.

We outline the ‘productivity paradox'; an economic argument that computers and information technology systems do not seem to pay for themselves. The productivity gains in companies that implement IT system do not seem to significant outweigh the cost of implementing the systems.

The productivity paradox is controversial and has exercised computer scientists for over a decade. We look at the argument from different angles; there are arguments that the paradox does not exist at all and it is merely a myopic use of statistics. There are compelling arguments that the paradox does exist and that it is caused by systems being designed without sufficient consideration for the users of those systems.

We then look at arguments for the cost effectiveness of making systems more usable as well as outlining some success stories.

Whereas the arguments for developing usable systems may be made explicit and quantifiable in a business context, technology impinges on our lives in ways which may be much less tangible. Badly designed systems can lead to anger and frustration for the users and may reinforce ‘technophobic' attitudes. Only by understanding humans better can such detrimental consequences be designed out of a system.

However, many of these issues are very subtle, take effect in the long term and cannot be easily quantified and costed. We look at how designers are (or maybe compelled to be) motivated to produce systems which are not detrimental to users in the absence of a cost argument.

We look at ethical issues in design, where users have rights not to suffer unusable technology and developers have responsibilities to respect those rights, as well as legal issues where usable systems may be required by law.

Technological systems are now extensively being put in place in situations where the failure of the system can result in severe injury and loss of life. Whereas ‘productivity' is central to cost effective systems, ‘reliability' is central to safety critical systems, where reliability describes and measures the likelihood that a system may fail with undesirable consequences.

Human operators control safety critical systems and are therefore, in effect, as much a part of a safety critical system as the technology. An understanding of human behaviour should lead to a system being designed which takes into account the behaviour of its operators and therefore is less likely to fail owing to ‘operator error'.

We also look at ethical issues of responsibility; the failure of safety critical systems may dramatically effect the lives of people completely external to the system. In many non-safety critical systems direct legal and ethical relationships can be established between developers, users of a system and consumers of the services provided by the system. Safety critical systems may impinge on a greater stage; if you are knocked down by a motorist driving a hire car whose ABS brakes are faulty, who is responsible? The driver for bad driving? The leasing company for hiring a defective car? The manufacturer? The sub-contractor who developed the ABS system for the manufacturer? Or yourself for crossing the road while knowing that there are cars out there with faulty brakes?

In many engineering fields ‘standards' are proposed which products should aim to satisfy. Products which conform to those standards are held to have attained an objective level of quality and are therefore more saleable. Standards for usability are immature, controversial and not widely used. Many developers claim that their products are ‘user friendly' but do not often explain what that claim means. Standards offer a way of objectively measuring a product for usability, and therefore a motivation for developers to make more usable products.

These four viewpoints on usability are not exclusive; safety critical system typically must meet cost requirements, etc.

Despite all the motivations for developing usable system we shall lay out, there is very little evidence of user centred design being successfully used in the commercial development of systems. We conclude this unit by suggesting some of the reasons for this apparent failure.