The MBIE Debacle – IT Failure Without a Fairy Tale Ending

VIEWS FROM OUTSIDE THE APIARY: IAN FLETCHER

The immigration minister says senior Ministry of Business Innovation and Employment (MBIE) staff misled her about a failed IT project costing $33million. Former top bureaucrat Ian Fletcher explains that such public-department projects, which can promise much, often suffer a sad ending, especially in New Zealand.

By Ian Fletcher

Immigration New Zealand hoped to improve their systems through a new IT system supplied by a private contractor, but were left with sunk costs.

The phrase “… and they all lived happily ever after…” has never been used around government IT projects. Yet the fairy tale “once upon a time…” promise is how they all start, with simplistic claims that the magic of IT and a cheap cloud computing “solution” will take cost, complexity and frustration out of some government service or other, and make it wonderful. And cheap. All that’s missing, it seems, is a fairy Godmother. There are plenty of evil cartoon characters, castles (mainly in the air) and simple plot-lines based on naïve hope, and a lot of magic taxpayer money from the bottomless well of plenty called “the Treasury”.

The stoush in recent weeks over the MBIE immigration biometric identification system has shown all these features (including the immigration minister as Sleeping Beauty, waking up to find her castle taken over by a bunch of goblins). The Public Service Commissioner seems to be cast as a hybrid of Shrek and Prince Charming. The show continues.

This is entertaining politics in an election year. But there are two big issues here: How we manage government IT projects, and how we deal with failure, especially in the public service.

Managing IT projects

Governments are naïve customers. They believe that government services (and government itself) can be made ‘better’ through digital systems. In some cases, this is true: if people want to get stuff done online, then that’s good. But government services are complex, and people’s needs are often hard to reduce to a set of algorithmic rules (which is how computers work). Sometimes, humans need humans to make decisions.

And governments sometimes don’t understand their own business: for example, the IRD works well enough if it collects roughly the right amount of tax from roughly the right people, roughly on time. Mistakes can be corrected later. In contrast, the Ministry for Social Development’s welfare system needs to get exactly the right money into the right account right on time. Errors really harm real people.

The immigration system is complex. It involves decisions about visas, so the right people are allowed in. But it’s a decision ahead of time, so it can be tolerant of system error, but not of errors in outcome (so we get the right people in at the end of the day).

Separately, there’s the border management system, so we know who has actually arrived. This is very sensitive – it has to let New Zealanders and Australians in easily, and apply other checks to visitors and migrants, all while sharing border information with others (like the police and the Australians). This was the area where MBIE’s project came unstuck. Others (like the EU) are struggling with biometric border systems too. Problems should have been expected.

Governments are naïve customers, especially when it comes to IT systems, says Ian Fletcher.

There is a second IT project weakness: relying on the private sector. Governments are poor customers because they don’t know enough about IT system design and operation, and they rely on vendors’ claims. Vendors have no interest in giving taxpayers value – they want to capture it for themselves. Governments also believe the hype – cloud computing is the answer, or AI is the answer. None of this is true – effective IT projects start with a really good grip on the business. There is a long list of failed public sector projects around the world, and the common factor is when the government hands over the keys to the vendor, and then wonders why the company cuts costs and fails to fix things.

The solution is to retain an in-house capacity to do some work and to provide the skills needed to judge others’ claims and performance. Otherwise, the fairy tale continues.

Is AI the solution? The best we can say is that the jury is out. Initial studies report real benefits in medical note-taking and associated administrative work. But other benefits are less clear-cut, especially as there is no accepted method of measuring public sector productivity. This isn’t a cop-out; it reflects the genuine complexity of public sector work, and the fact that it often can’t just be broken down into a series of tasks, like a factory.

Dealing with Failure

The MBIE project failed. Then it seems it was kept going in a zombie state, and then finally killed off before Ministers and others were really aware. The Minister has accused officials of a cover-up, and of deliberately avoiding scrutiny by limiting expenditure levels to fall below the thresholds for escalation. It’s clear something has gone wrong.

Things in government go wrong a lot, and (in New Zealand especially) that’s often treated as a surprise, because we have a real problem with everyone thinking the same (or pretending to), and everyone imagining that we’re special and things will always go well (or pretending to). So, we handle failure badly when we’re forced to face it. It’s the Emperor’s New Clothes fairy tale, acted out.

Why’s That?

Two big reasons: first, we are over-centralised. Chief executives are given a lot of authority, but then hemmed in by compliance rules, a lot of mark-your-homework behaviour from the Public Service Commission, and a high degree of personal exposure – so ducking for cover is a habit taught from the top. Ministers don’t – or won’t – back officials.

Secondly, there’s a real reluctance to think for ourselves. The easiest way to get things approved in Wellington is to claim it’s what’s done overseas, whether or not it’s a good idea (or even true). And of course, overseas IT companies win both ways – they can pull the wool over departmental eyes, and then claim that’s what other countries do. The sales pitches from big US IT providers follow this script.

Finally, recent research shows that believing in business jargon (including IT jargon) is a sign of limited intellectual ability. One UK government senior management course I attended made us explain complex policies to a class of seven-year olds. It certainly defeated the jargon. And it really did end happily.

Ian Fletcher is a former head of New Zealand’s security agency, the GCSB, chief executive of the UK Patents Office, free trade negotiator with the European Commission and biosecurity CE for the Queensland government. These days he is a commercial flower grower in the Wairarapa and consultant to the apiculture industry with NZ Beekeeping Inc.