Nobody had to mean it. The informal rules won over the formal ones and so decided who held power, and it wasn't the Minister, Natalia Albert says of Immigration NZ's shambolic tech project

By Natalia Albert*

Erica Stanford is furious. Officials at Immigration NZ ran a biometrics project for seven years, burned through $31 million, delivered nothing, and along the way appear to have misled at least three ministers, split a project in two to dodge Cabinet, and tried to bury the costs in an unrelated paper. Objectively problematic.

When Stanford finally got the independent review, she called it "almost as bad as it gets." The Public Service Commissioner is now promising to leave no stone unturned. Winston Peters wants prison. Andrew Little remembers some of it. Kris Faafoi is nowhere to be seen. The Ministry of Business, Innovation & Employment (MBIE) has a new chief executive who now must wade through years of executive and institutional decisions to figure out what the f*ck happened. This is bad. 

If even half of what Stanford described to the select committee holds up, some people behaved appallingly, and they should answer for it. But here's where I want to slow the whole thing down, because the entire debate has collapsed into a single question, and it's the wrong one.

The question everyone's asking is: did they do it on purpose? Stanford says deliberate. She used the words "creative accounting," "deception by omission," a "trifecta of terrible things."

MBIE's chief executive Nic Blakeley reaches instead for "optimism bias," the idea that people who'd sunk years into a project kept telling themselves, and their minister, that it would come right. The Public Service Commissioner's review will, we're told, settle which of these is true.

I don't think it will. And I don't think it matters as much as everyone assumes. Because underneath the question of intent sits a bigger one, we're not asking: what do these officials think their job is? In a liberal democracy, public servants serve the government of the day. They advise, they implement, they tell ministers the truth even when it's ugly. What they don't do is decide for themselves which projects are worth protecting, then manage the elected minister's information so the decision stays theirs. That's not a clerical failure. It's a misunderstanding of who holds power and who answers to whom.

There was a rule: if a project's whole-of-life cost goes over $35 million, it needs Cabinet approval. A clean, sensible formal rule. The kind of guardrail you'd design specifically to stop officials quietly committing taxpayers to expensive things without political sign-off.

Now look at what the independent review found. When the project crept towards that threshold, officials split it into two. The review's words, not Stanford's: the split was "largely driven by efforts to avoid exceeding the Cabinet-level funding threshold." They didn't break the $35m rule. They obeyed it by making sure no single piece of the project ever technically crossed the political threshold.

This is the part the villain-hunt misses entirely. The $35m threshold functioned as a target to avoid, not a target to respect. It told everyone involved exactly where the bright line was, and the entirely predictable human response to a bright line you don't want to cross is to arrange your affairs, so you stay just underneath it. Dumb. 

This is a well understood political science phenomenon. Formal vs Informal rules.  Formal rules, the ones written in glossy ministerial paper, on websites, in annual reports, into legislation, the thresholds and signoffs and Cabinet papers.

Formal rules also never operate alone. They sit on top and amongst informal rules: the unwritten norms about what you do outside the formal ones to sidestep the ones that get in the way, what gets escalated, what gets quietly handled, what your boss wants to hear.

Gretchen Helmke and Steven Levitsky, who did the foundational work mapping this, point out that informal norms fill the gaps left by formal ones, they actively compete with them, producing outcomes the formal rules were specifically designed to prevent. 

They call these "competing informal institutions," and Immigration NZ is a textbook case. When the two line up, the system works. When they pull against each other, the informal rules usually win, because they're the ones that govern day-to-day behaviour and carry the real social consequences.

At Immigration NZ, the formal rule said get Cabinet approval over $35m. The informal rule, the one that ran the building, said don't bring bad news up the chain, protect the project, keep the Minister comfortable. We know the informal rule was real because the review found people who raised concerns were moved off the project. That's not optimism bias. Optimism bias doesn't reassign your colleagues. But it's also not necessarily a criminal conspiracy. It's a culture doing what its incentives told it to do.

And this is why I think Stanford, for all that she's substantively right, is reaching for a frame that lets the system off the hook. "Deliberate deception" locates the problem in a few bad individuals. Find them, sack them, and the integrity of the public service is restored. It's a satisfying story. It's also the story that requires the least change. 

Because if it's deliberate deception by a handful of people, you don't have to ask why three successive ministers, across two governments, all got managed the same way. You don't have to ask why Andrew Little was being told everything was "hunky-dory" while it clearly wasn't.

You don't have to ask why RNZ's Gill Bonnett was reporting red flags on this project back in 2024 and the official line was that the risk profile had "not worsened." You don't have to ask whether the $35m threshold itself, the supposed safeguard, was part of the problem.

A threshold that tells people exactly where the line is will always be gamed by someone. The question is whether your system notices.

Here's the pre-emptive concession, because someone will raise it: am I excusing the officials? No. People made choices. Moving colleagues who raised concerns is a choice. Sitting on a finished review for two months is a choice.

Telling a brand-new minister the project was "sound and robust" when the actual assessment doubted it would deliver "at all" is, at absolute minimum, a choice not to be honest. Institutional analysis isn't a get-out-of-jail card. It's an explanation of why those choices felt available, even reasonable, to the people making them. Individuals are accountable. But if you only hold individuals accountable, you'll do the whole thing again in five years with different individuals.

The genuinely uncomfortable part, the part that should worry you more than any single villain, is the bipartisan one. This wasn't done to a National minister by a hostile bureaucracy. It was done to Labour ministers too. Little has effectively admitted his own government was kept in the dark, even as he sympathises with Stanford. That should kill the left-right framing dead, though it won't, because an election year needs a story and "the institution malfunctioned regardless of who was in charge" doesn't fit on a placard.

And the structure made it easy. In 2022, partway through the project's life, it was shifted out of Immigration NZ's hands and into MBIE's digital, data and insights group as part of a restructure. On paper, a tidy reallocation. In practice, the kind of move that quietly dissolves accountability: the people who started the project no longer owned it, and the people who inherited it never really did either.

When something belongs to everyone, it belongs to no one. This is the bit institutionalists would flag in a heartbeat, because formal structure isn't neutral. Draw the org chart so that responsibility falls between two boxes, and you have built a place for bad news to get lost. Nobody must hide it. The structure hides it for them.

So, what would actually fix this? Not a list of sackings, though some may be warranted. The honest answer is the unsatisfying one: you have to design rules that assume people will respond to incentives, because they will. A threshold that everyone can see is a threshold someone will engineer around.

A culture where bad news travels down instead of up will produce misled ministers no matter how many integrity reviews you run after the fact. Nicola Willis said the right thing this week: do not cover up bad news, send it up to the Minister's desk. The trouble is you can't instruct a culture into existence. You have to build a system were sending bad news up is the path of least resistance, not an act of career-risking courage.

Brian Roche says he'll leave no stone unturned. I believe him. I just hope that when he turns them over, he's looking for more than names. Because the names will change. The threshold, the silos, the incentive to manage your minister rather than inform them: those are still standing. And they'll be there for whoever runs the next doomed seven-year project.

*Natalia Albert is a political scientist living in Wellington exploring how to govern divided societies in diverse, liberal democracies, with a focus on New Zealand politics. She writes weekly on her Substack, Less Certain. Albert stood as a TOP candidate in the 2023 election.