Databases Reference
In-Depth Information
poor usability often means people maintain parallel information systems. Also, if it is difficult to
enter pertinent information, they may simply not bother to do so. Even though the problems are
with superficial usability, political and social context are fundamental to the design of the data
structure; in all cases, when people in an organisation keep parallel systems, the centrally held 'official'
organizational data becomes unreliable and inconsistent with reality. Paradoxically, the attempt in
the database to maintain data integrity, in the narrow sense of being a self-consistent model, actually
works against
real
integrity so that the data held begin to diverge from the things in the real world
they are intended to represent.
However, things can be much worse. In 1992 the London Ambulance installed a new
computer-aided dispatch; although it was intended to be a new sleek automated system, it be-
came a potentially fatal disaster (
Sommerville, I.
,
2004
). As soon as it was deployed, it became clear
that, rather than streamlining dispatch, delays increased, and a short while after deployment, the
entire system collapsed leading to an emergency re-introduction of the old manual system. There
were numerous failures but not the least amongst these was a failure to take into account the com-
plex human and organizational factors involved in the dispatch process. Although there was some
debate, no deaths were unambiguously attributed to the incident, this was only by good fortune.
Poor usability can kill.
Happily, the costs of usability failure are not always so severe, and in practice, one must strike a
balance between the efforts spent making a system more usable and the benefits of the improvements.
In fact, there are two stages to this. First, find potential problems or potential things to
improve; second, decide whether it is worth implementing a solution. For the latter, it is possible, in
principle, to do a simple cost-benefit trade-off. Figure
1.1
shows this with the severity of a usability
problem on the left axis and the cost to fix on the bottom (focusing here on problems, a similar
analysis applies to potential improvements).
fix!
damage limitation
low cost ways to
reduce imipact
severity of
usability problem
serious
(a)
(b)
?
(c)
(d)
?
live with
minor
low
high
cost to fix
masks other problems - fix!
Figure 1.1:
Cost-benefit trade-off for usability problems.
