NZ First and Act are both on the trail of more information about the costs and alleged instability of Swiftt, the venerable social welfare information system that critics say should have been replaced long ago.
Meanwhile, at least one software supplier is prepared to stick its neck out and budget for a complete replacement. Helen Robinson of CRM software vendor Pivotal Systems says the essential functions could mostly be supplied “out of the box” by a standard customer relationship management system and the whole job done for less than $10 million.
This compares with the Ministry of Social Development’s own $78 million to $120 million estimate in 2002, and predictions as high as $500 million from a source previously closely associated with the system; someone MSD prefers to describe as “a disaffected ex-employee”.
Social development minister Steve Maharey, replying in the House to a question from NZ First leader Winston Peters, has said Swiftt is still “reliable”. Hardware has already been upgraded from a Unisys Clearpath to a Unisys Libra mainframe, and the software is set for incremental upgrade, Maharey says.
“There is a modular plan to upgrade software through the system. A governance process has been in place since we have been in government. There is absolutely no doubt that this system will deliver for beneficiaries and superannuitants.”
However, NZ First researchers claim to have uncovered a near collapse of the system, said to have happened within the last two months. The party is seeking further information on this incident.
Act’s Muriel Newman says there was apparently a rethink of a major upgrade in the late 1990s, after the police’s Incis project was terminated. The Social Services Committee, hearing estimates for that slice of the budget in 1997, was told that $120 million was to be spent on renovating Swifft over the following three or four years, Newman says.
“After Incis, that $120 million suddenly seemed to disappear. In response to more recent questions, we have only heard much smaller sums in single-figure millions or tens of millions.”
She supplied Computerworld with the written answer to a question asked earlier this year about expenditure on Swiftt. This summarises expenditure funded by government to satisfy new policy initiatives, for which the highest figure is $10,063,311 for 2001/2, and “system enhancements and application support costs”, funded from MSD’s own budget, in the region of $2 to $3 million a year, with the highest being $4,968,761 in 2002/3.
Post-election briefing documents, in 2002, estimated the total cost of enhancing or replacing Swiftt at between $78 million for the upgrade option and $120 million for replacement.
Unisys will not be drawn into estimating a cost for total replacement. Gen-i chief Garth Biggs says he does not know enough about the system to venture a comment, much less an indicative price. From his experience in the airline industry (emphasising that he is not referring specifically to Air New Zealand, where he was CIO) he says he has seen reservations systems credibly costed in the $100 million region.
“Government departments always take the ‘big bang’ approach and presume that no existing tools are available on which to base a system,” says Pivotal’s Robinson.
“A typical government department would laugh if you said you could base a system to meet their needs on an out-of-the-box CRM package. But that’s what [the social welfare function] is about. They’re customers, and good CRM is about people managing their relationships with people.”
The relationships are complex and many-to-many, but that is within the capability of modern package software, she says.
“You specify who the people are and how they relate and half the work’s done.”
Robinson claims enough knowledge of Swiftt’s workings to suggest her less-than-$10 million estimate is reasonable, but will not say how she came by that knowledge.
SolNet’s Mark Botherway says a “notional short-list” of possible software sources that MSD is said to be considering for a future system leans heavily in the CRM direction.
Another industry source suggests payroll might be a better off-the-shelf starting point.
“Instead of having all these add-on benefits, you sweep them all up into a system that says ‘you’re a grade 5 beneficiary, so you get this much’; then a modern payroll and [human resources] system should handle it.”
Act’s Newman, in questions put last week, asked for further breakdown of the spending figures between “system enhancement” and “application support” as well as the analysis of individual policy initiatives to which Swiftt enhancements apply. Replies are expected late this week or next.
She also asked whether MSD consulted with the e-government unit over a failed proposal to “e-enable” Swiftt, allowing more information to be retrieved online.
Peters and Newman have both asked what documentation exists for the system, there being some doubt whether documentation is complete. Peters has asked how the system can be proved to be stable, as Maharey says it is, without such documentation.