"So you get an error, and you say: 'Well, I think this is their error, but maybe its mine, and I don't really know. And I don't have any way of talking about it, because I can't speak to a developer. None of the people that are managing the project know about the technical aspects of things.'

"So critically - and this is the salient point - the real cost of development is an order of magnitude higher than it ought to be because you have to test everything many times, for every single version of the Customs system, because it's changing, it's unstable, and the whole nature of the thing is still under development while you're trying to build to it."

The result, White claims, is that developers are providing an outsourced testing resource for Customs that they never contracted to provide and that they are certainly not being paid for: White claims his company is down half a million dollars in terms of the cost of delivery.

White nevertheless speaks highly of Harrison, calling him "credible" and "balanced" in his assessments of the program and its problems, while positing that the Customs CIO is clearly constrained in what he can say. White also acknowledges the size and complexity of the project, the fact that many IT projects run over budget and over time, and that journalists have tended to sensationalize the story.

But you have to look at the situation from the most abstract possible point of view, he says, and that abstract view is simply that Customs needs to be able to develop and deliver systems that are within reasonable bounds of quality and stability.

"Look, this was always destined to be hard to do," White says. "You can't get down on Customs about the fact that the project is going to cost 'x' or it's going to take 'y' to do. What you can do is get down on them about the fact that they were so insistent it was going to be done in 'x' and cost 'y' . . . They absolutely made those commitments to industry at that time in writing, and repeatedly.

"There are some views in Customs that nothing is wrong and everything is going swimmingly and this is exactly as they planned. I think that's probably a very unreasonable view because I don't think anybody who has a historical understanding of what's gone on can say that this is what was promised or even in the remotest sense understood to be what the process was going to be," he says. "No one outside Customs can say what the detail of the problems are, but if you see a house with smoke pouring out of all the windows, you pretty much know that there is a fire inside."

In fairness, though, White adds that the attitude of Customs is observably undergoing major changes. The agency is now actively working on a much better test regime, he says.

Megaproject

Finally there is the issue of the size of the project. Some critics have claimed the real reason the development is so massive is that senior management is determined to prevent any moves by politicians to separate the agency's major functions: the control of the movement of goods across the border and the expedition of that movement of goods.

Whether those claims are right or not in ACS's case, here too, Hall has something to say, under the heading "The Megaproject: If some is good, and more is better, then too much must be just right".

" . . . the public sector's bureaucratic funding and approval processes encourage counter-productive aggregation of smaller projects into megaprojects because people dread 'multiple trips to the well'," Hall writes. "Breaking projects into discreet subprojects would reduce risk but would also entail shepherding multiple business cases through the approval and funding process, thus increasing the risk that some projects would not make it safely through the bureaucratic morass.

" . . . Further, to obtain funding in the public sector, the business case for the megaproject must present cost, schedule and scope factors with unflinching confidence. Only in the federal and state arenas would anyone attempt to fully specify the cost, schedule and scope of a $50 million to $1 billion IT project at the beginning and expect that vendors could develop credible fixed-price/fixed-schedule bids for something that large and complex."

Enough said.