Resolving on-line reading issues in Software Inspection
Traditionally, most subject-based experiments agree that, in general, human performance
when reading from the screen is inferior when compared to paper based reading. However,
recent experiments have suggested that improvements in screen technology and
innovative approaches to screen layout can reduce and could eventually eliminate
these problems. This deficiency is compounded by current software inspection
support systems and the inspection processes. Unlike traditional reading tasks,
an inspection process requires the reader to view multi-documents causing massive
screen space issues and potentially large context switching overheads if the
reader has to continuously manage their screen space. In addition the support
system, may well use more space for administrative (to manage the support system)
issues and will fail to match the convenient annotational and note-taking style
of traditional pen and paper systems. Hence, the obvious conclusion is that
current screen and keyboard infrastructures utilised by inspectors and their
support systems are deficient in supporting this extended reading process. This
deficiency will also exist in other software engineering activities; for example,
many authors estimate that code will be read nine times more often than written
or modified.
Hence, as part of the on-going research within the ASSIST project, the team would
like to explore the use of new large screen plasma displays to reduce or eliminate
deficiencies in on-screen reading within the context of a software inspection
support system. These displays offer large, high quality, display areas allowing
the increased resolution of the documents under inspection. Although these screens
offer a significant advantage over traditional technology, they are still currently
insufficient to meet the demands of a multi-document inspection technology.
Hence, it is envisaged that a system would consist of four displays mounted
in a structure akin to a large ‘draughtsman board’. All of the displays will
be linked into the single machine running the support system. Finally, to limit
the use of the traditional keyboard and mouse arrangement these screens will
be interactive, by adding touch-sensitive technology to the plasma display.
This will allow the inspector to touch and drag information around the display
area and annotation directly on to the screen with a ‘virtual pen’ without resorting
to interaction with the keyboard or mouse. Although this technology is currently
relatively expensive, it is believed that its price will quickly drop once it
enters into the mainstream market place. When we compare it to the salaries
of software professionals, it is easy to see that efficiency gains by the introduction
of this technology will quickly justify itself.
It is intended to adapt the current version of ASSIST to attempt fully utilise
the new facilities offered by the introduction of this additional hardware.
Unfortunately, little is understood about the utilisation of such facilities.
In fact, little research exists on the impact of non-traditional layout and
structures on reading tasks. Perhaps, the most comprehensive study is by Dyson
et al. who, via a series of experiments on narrative text, challenge
many of the assumptions about the impact of layout and structure. This project
plans to utilise this work as a starting point. Hence, it is planned to run
a number of subject-based experiments to attempt to quantify the impact of the
new technology over the existing technology. It is believed that such hard
quantitative results will help to convince industry of the benefits of adopting
this new technology; and the results will allow the construction of precise
cost-benefit models to inform companies of the potential savings when undertaking
inspections on a regular basis. Finally, it is hoped to extend these studies
in two directions: (1) by looking at further software engineering tasks and
processes to assess the impact within the entire software engineering life cycle;
and (2) by investigating the possible addition of other media types and peripheral
devices, such as audio and pen-based components.