Guest Post:  Seeing the System: Why Government Struggles to Fix What It Can’t See 

A guest post by Chris Scott:

 Most New Zealanders can picture a family farm. Not the romantic postcard version, but the real thing: a few hundred acres, a shed full of tools, and a family that has been arguing about how to run the place for as long as anyone can remember. Anyone who’s ever tried to run a farm with family members knows that politics isn’t confined to Parliament. 

Beneath the noise, it’s a system. It has inputs and outputs, constraints and feedback loops. Some constraints come from the land itself — soil, water, weather. Others arrive in the form of rules someone in Wellington has decided are important. Some of those rules feel pointless; others quietly save you from a mistake you didn’t know you were about to make. 

A farm like this is small enough that you can see the whole thing at once. You know where the water comes from, where the money goes, which paddocks are struggling, which machines are on their last legs, and which decisions are likely to start an argument at the dinner table. You don’t need a consultant to tell you where the bottlenecks are. You can feel them. 

Government is the same kind of system, just scaled up until no one can see the whole thing anymore. The incentives are still there. The arguments are still there. The constraints are still there. But the visibility is gone. The people running the system are buried inside it, and the people affected by it only ever see their corner. 

You can see the consequences in projects like the Auckland Harbour tunnel. It was conceived, consulted on, modelled, costed, debated, redesigned, and eventually abandoned — after years of work and hundreds of millions of dollars spent. And the bottleneck it was meant to solve is still there. 

From a systems perspective, this isn’t surprising. The project wasn’t just a tunnel. It was a tangle of subsystems: 

transport modelling environmental constraints local and central government politics procurement rules engineering risk public consultation budget and election cycles regulatory approvals inter-agency coordination 

Each of these has its own incentives, timelines, and failure modes. And no one — not ministers, not agencies, not consultants — ever gets the full 10,000-metre view. Everyone sees their slice. No one sees the whole machine. 

This is why the project could be technically feasible, economically justified, politically supported, and still fail. Not because the people involved were incompetent, but because the system they were working inside was never decomposed, modelled, or redesigned as a whole. 

If you applied a modern decomposition approach, you wouldn’t begin with the route or the price tag. You’d begin by separating the system into parts that can be understood on their own terms. 

Once separated, the shape of the problem changes. 

You can see where delays accumulate. You can see where incentives clash. You can see where information gets lost between agencies. You can see which constraints are real and which are inherited from older decisions that no longer make sense. 

And once you can see the system, you can model it. 

You can test scenarios before committing to them. You can identify points of fragility. You can distinguish structural constraints from artefacts of process. 

This is, in essence, what modern AI and software tools enable. 

Inside large software systems — with millions of moving parts, legacy decisions, and hidden dependencies — these tools are used to map relationships, trace flows, and surface bottlenecks. For example, engineers can simulate how a single change propagates through a system before deploying it, avoiding costly failures downstream. 

It’s not magic. It’s visibility. 

And it’s visibility our government doesn’t currently have — at a time when the complexity of the problems we face is only increasing. 

If we’d had that kind of visibility for the tunnel, we might still have decided not to build it. But we wouldn’t have spent years and hundreds of millions of dollars discovering that fact the slow way. 

A family farm can survive on intuition because the system is small enough to see. A government can’t. The problems we’re trying to solve now are too large, too interdependent, too full of hidden constraints. 

If we want a government that works, we need to give it the ability to see itself. 

Right now, it can’t. 

Postscript: On AI’s role 

I drafted this piece myself, but I used an AI collaborator to help refine the structure, test the logic, and tighten the language. The ideas are mine; the AI helped expose weak points and iterate faster. It didn’t write the argument — it sharpened it. 

The post Guest Post:  Seeing the System: Why Government Struggles to Fix What It Can’t See  first appeared on Kiwiblog.