The best map sometimes leaves things out

GIS people are naturally rewarded for adding information. Another layer appears, another dataset is joined, another set of points is mapped and the system becomes richer. That is usually presented as progress. Māori GIS taught me that a good map is sometimes defined just as much by what it deliberately leaves out.

That is not the same as hiding an error or producing an incomplete map through carelessness. It is a decision about kaupapa, authority and responsibility. Some information is too sensitive for a public map. Some locations should be generalised. Some details belong in an internal system rather than a website. Some knowledge may not need to become digital at all. The fact that GIS can map something does not mean it should.

This became obvious during claimant mapping. GIS teams could hold information supplied for very specific purposes. A site might be mapped to support research or a hearing. A historical map might be loaned to answer one question. A claimant group might provide detailed information that was never intended to become a general-purpose dataset. Once the information was digital, however, the software made reuse extremely easy. A layer could be copied into another project in seconds. The technical friction disappeared while the relationships and responsibilities around the information remained.

That distinction is one of the strongest influences te ao Māori has had on the way I now think about mapping, although I do not claim to speak from within that worldview. Conventional GIS tends to treat information as an object. Once it has been digitised it can be stored, queried, copied and published. Māori mapping work repeatedly reminded me that information can remain connected to people, whenua, whakapapa and the circumstances in which it was shared. The file does not become detached from those relationships simply because it now has a coordinate system.

That is where possession and authority part company. The GIS operator may have the file without having the right to publish it. An organisation may legitimately maintain detailed information internally while deciding that the public version should contain much less. The people with authority over the information may also decide that the map should not exist at all.

This is not a failure of openness. It can be an expression of rangatiratanga.

Modern web mapping makes the decision more important because publication can be effectively permanent. A layer can be copied, cached, downloaded or reproduced elsewhere. Removing the original application later does not necessarily remove all copies of the information. A sensible publication workflow therefore starts by asking what the kaupapa actually requires and who has authority to decide what is shown.

If the purpose of a public map is to show the general location of an environmental project, exact coordinates for sensitive places may add nothing useful. If a StoryMap explains historical change, an approved generalised layer may communicate the story without exposing the complete source dataset. If the public only needs to understand the catchment, there may be no reason to publish every point collected inside it.

Cartographers have always generalised information for scale and readability. Roads are simplified, minor features disappear and labels are selected. Māori GIS adds another reason to generalise: sometimes exactness is not appropriate. The best representation may be a broader area, an intentionally incomplete attribute set or a public derivative prepared from a much richer internal source.

The key is that the decision should not be made by the software. It should be made deliberately by the people who hold the appropriate authority. The technical specialist can explain the implications, prepare different options and show what may be inferred from a map, but that does not make the specialist the person who decides what knowledge should become public.

This is also where kaitiakitanga has influenced my thinking. I do not use the word as a direct substitute for “data stewardship”. The point is responsibility. If information has been entrusted to an organisation or practitioner, holding it creates obligations. Those obligations can include protecting context, limiting unnecessary copies, maintaining appropriate access and considering what happens when today's project team is gone.

A related question is whakapapa. A GIS layer may have its own technical provenance, but the knowledge behind it can have a much deeper lineage. A point on a map may relate to generations of occupation, memory or kōrero. The point is not the whakapapa. It is only a spatial representation associated with it. That difference should make the mapper cautious about assuming that more detailed digital representation is always better.

Sometimes the right answer is to preserve the relationship without publishing the detail.

The same principle is useful for StoryMaps and other public storytelling tools. A StoryMap can use approved web layers without becoming the system of record. The story can show what is needed for the public narrative while authoritative or restricted information remains somewhere else. This separation gives communities more control over what is communicated and helps prevent a public storytelling application from quietly becoming the place where all source knowledge ends up.

The same principle now matters with artificial intelligence. There is growing pressure to feed information into systems because doing so might produce a useful answer. But an AI tool does not need every real dataset in order to explain a GIS operation, design a schema or help write a workflow. Sample or synthetic information may be enough. Sometimes the most capable workflow is the one that deliberately gives the tool less.

This is not anti-open-data. Public government data has transformed Māori GIS capability by making basemaps, imagery, elevation, property information and many other resources easier to access. Open data is extremely valuable where the information is genuinely public. The problem is turning openness into an automatic destination for every dataset.

A Māori framing offers a more relational question. What is the right way for this information to be held and used in relation to these people, this whenua and this kaupapa? Sometimes the answer is public. Sometimes it is shared within a defined group. Sometimes it is intentionally generalised. Sometimes it remains local. Sometimes it should remain in kōrero, memory or another form rather than becoming another GIS layer.

That last option is easy for GIS practitioners to overlook because our professional instinct is to map. But not every form of knowledge is improved by conversion into points and polygons. The right not to map something is part of genuine control over information.

The best GIS practitioner is therefore not the person who can get the most information onto the screen. It is the person who understands enough about purpose, authority and responsibility to know what belongs there, what should be shown differently and what should be left alone.

For a deeper discussion of this approach, see We will never ask you for your data and State of Unknown.