This presentation revisits a recent critical review of the Novopay project presented at the International Conference on Global Software Engineering in July 2013. That study which adopted a ‘critical evaluative’ stance, applied dilemma analysis to review the implementation of a nationwide payroll system responsible for the payment of some 110,000 teachers and education sector staff [1]. The project stands as a notable example of a large New Zealand Government Business Process Outsourcing project involving a nearshore provider. It is now well known that the project had become so troubled that a Government Minister has been assigned responsibility for troubleshooting it, a portfolio which he still holds. The Australian company who won the contract for customizing and implementing the Novopay system, took over from Datacom an existing New Zealand service provider. While commonalities in business practice and cultural awareness might be expected in projects spanning a relatively small temporal and geographical distance, in this project that did not appear to be the case.

Obvious parallels between the latest cutover and the previous two cutovers in 1989 [2] and 1996 [3] are drawn, as are comparisons with the recent Queensland Government health payroll debacle [4]. The latter project had its roots of failure in a different political context and origins, but presented a similar case of a transition from an old yet functioning but very complex payroll system, and resulted in a commission of enquiry. It is notable that after the experiences of 1996 the then Chairman of Datacom had opined, “Lessons learned about the rate of change possible in large systems will never be forgotten by those involved”, and “This big bang approach is a recipe for failure” [3]. But evidently memories in such major, costly and high risk endeavours are short, and the ability to transfer such knowledge to other (even nearshore) contexts seems limited! Practitioners do not seem to read the academic literature either, where many of the pitfalls in not addressing concerns from a multi-stakeholder perspective have been outlined, together with strategies for addressing them [2, 3, 5, 6, 7].

It is clear that large complex software transitions of this sort are challenging for all parties involved, and beset with dilemmas. So how might we better learn from them? In the study reviewed here the potential for dilemma analysis [8] to highlight risks and stakeholder impacts is illustrated. To our knowledge this is the first such use of dilemma analysis (which has its origins as a methodology for educational research) in a global software engineering study. Empirical analysis, in this case, of publicly available data, has adopted an evaluative-critical approach. We argue that this methodological approach to such projects can usefully highlight tensions and barriers to satisfactory project outcomes. The study starkly illustrates the extent to which large software projects are beset with dilemmas, which must be navigated by the parties involved. This work is the subject of an on-going doctoral study by Bilal Raza at Auckland University of Technology.