Another report about why IT projects fail

A few days ago, the Royal Academy of Engineering and the British Computer Society published 'The Challenges of Complex IT Projects'. There's considerable criticism of current practice: in general, those in the sector know what best practice is, but fail to follow it, according to the report.

Usability gets a few mentions (2 to be precise), as do the poor old users. Though user-centred design is not explicitly mentioned, there is some criticism of the 'waterfall', or sequential approach to projects, and support for iterative development.

Fred Brooks' Mythical Man-Month, originally published in 1975, gets quite a few supportive mentions. As the report states, "Despite its age it contains observations and advice of great relevance to current projects, which in itself is telling."

By the way, there's an improved anniversary edition of MMM, containing a number of later essays, including the widely cited 'No Silver Bullet: Essence and accidents of software engineering'.

All too often, a major problem is the decision: what to design? I feel the issue of requirements gathering gets inadequate treatment in the report. It was, after all, Fred Brooks himself who wrote, "The hardest single part of building a software system is deciding what to build...No other part of the work so cripples the resulting system if done wrong. No other part is more difficult to rectify later".

One recent UK example of failing to research user requirements was the Criminal Records Bureau system, designed to provide obligatory criminal record checks for prospective teachers, social workers and many more employment groups. Designed by the contractor as a call centre-and-website to deal with individual applications, only as implementation was imminent - and the marketing roadshow in full swing - did developers talk to potential users of the system...and accidentally discover that the vast majority planned to apply for checks in writing (probably to cover their backs in this legal minefield). Nor had developers figured out that applications from user locations would be submitted not individually, but in groups (mostly because every applicant or interviewee for a particular job would need to be screened at precisely the same time - say, by the headteacher or other hirer). Some things seem so obvious when you find out about them...