Semantic Werks

I’ve tried to resist being swayed by the ’softer’ side of software science: business practices, management science, process improvement, etc., all of which I feel are challenging to evaluate scientifically.

However, a recent paper,  A Replicated Survey of IT Software Project Failures, suggests that if we want software projects to succeed in IT, than these aspects are perhaps the best place to focus our attention.The paper is a refreshing improvement on the Standish reports.

Of the top five reasons a project was cancelled, three were management-related: Senior management not sufficiently involved (33%), too many requirements and scope changes (33%), and lack of necessary management skills (28%). The other two were project over budget (which presumably has many possible causes), and a lack of necessary technical skills (22%).

As software engineering researchers, I would argue that a lot of our work is focused on the last reason (technical skills) or less important causes (e.g. technology problems). There is a large research community dedicated to requirements research, but arguably this community is seen as less important (compare, for example, papers in ICSE 2008: of  103 papers, only 6 were requirements related).

These results also suggest what I’ve suspected: the choice of technology (C#, Java, SQLServer, AJAX, etc) is less important than getting a good team together, with properly scoped requirements and a sound leadership vision. As Karl Pilkington would say, the rest is just ‘pfaffing about’.

Thanks to Jorge and Lorin Hochstein for the pointer.

This entry was posted on Saturday, September 27th, 2008 at 19:35 and is filed under General.You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.