State of Unknown: what Māori claimant mapping taught me about precision, uncertainty and authority
GIS has a habit of making uncertainty look tidy. Put an old plan on a screen, add a modern cadastral layer, draw a line and label it, and suddenly the result looks as though somebody has settled the matter. The line is crisp, the coordinates are precise, the parcel beneath it is authoritative and the aerial image may be sharp enough to see fences and buildings. Everything about the screen tells the viewer that the answer has been found.
A great deal of Māori claimant mapping taught me the opposite. Sometimes the most accurate result is to admit that the answer is not known. Historical land blocks do not always line up cleanly with modern parcels. Rivers move. Surveyors make assumptions. Names change. Plans disagree. Natural features disappear. Areas recorded on paper do not always agree with areas calculated from geometry. Some source plans are missing altogether. The software is always willing to draw a line, but that does not mean the evidence is willing to support one.
In one of the Northland mapping methodologies we had a very useful description for this: State of Unknown. It was not grand terminology. It simply meant that the available information did not justify pretending there was one certain answer. Over time I came to think of it as one of the most useful professional disciplines GIS can offer, particularly in Māori GIS, where whenua is not merely a parcel to be measured and knowledge of place does not become more authoritative simply because it has been converted into coordinates.
I am not Māori and I do not claim to speak from within te ao Māori. What I can describe is what repeated work alongside Māori claimants, practitioners, researchers and communities taught me about the limits of a purely technical way of looking at land and information. GIS tends to begin with objects: points, lines, polygons, parcels and layers. Much of the Māori mapping work I encountered began somewhere more relational, with people, whenua, whakapapa, history, responsibility and the authority to say what a place means. The map could help represent parts of those relationships, but it could never become the relationship itself.
The convincing map
I became involved in claimant mapping in the mid-2000s, by which time digital GIS was already extending earlier cartographic methods used in Treaty research and claimant work. Historical blocks were being reconstructed, Crown transactions mapped, place names corrected and research information brought into spatial systems that allowed layers of history to be seen together. My own role was practical. I worked with source plans, scanned maps, cadastral information, claimant research, imagery, map books and hearing material, often taking information created for completely different purposes and trying to place it into one spatial framework.
A nineteenth-century survey plan, a modern parcel layer, an oral-history reference, a Public Works plan and a current aerial photograph might all contribute to the same question, but they were not the same kind of evidence and they did not carry the same authority. Nor were they simply separate technical inputs. They sat within different histories and relationships. A government plan might show how the state recorded a transaction. A claimant map might carry knowledge maintained within whānau or hapū. A place name might connect to whakapapa and memory in ways that an official label did not. Putting everything on one screen was useful, but the screen could easily hide those differences.
That was one of the first important lessons for me. GIS is excellent at combining information, but it is equally good at flattening distinctions between sources. A polygon generated from an uncertain old plan looks just as neat as a modern cadastral parcel. A point supplied through local knowledge looks visually identical to a point downloaded from an official dataset. If the map maker does not carry the distinctions into the data model, metadata, symbology and explanation, the finished map can imply a certainty or authority that never existed.
The most dangerous map is not necessarily the inaccurate one. Sometimes it is the beautifully designed one whose uncertainty, relationships and provenance have disappeared.
When lines disagree
Historical boundaries are particularly good at resisting neat answers. An old block might be described using a river bend, ridge, stream junction or another natural feature. A later survey may interpret the same feature differently. The river may have moved. Another plan may show a different relationship. A modern cadastral layer can provide a precise contemporary legal geometry, but it does not automatically tell you what an earlier surveyor meant or how the whenua was understood before that cadastral system existed.
The technical temptation is strong because GIS gives you many ways to make things fit. You can georeference the plan differently, choose new control points, stretch the image, snap reconstructed boundaries to modern parcels or adjust geometry until the result looks sensible. In many ordinary mapping tasks those tools are exactly what you should use. In historical reconstruction they can also erase the problem you are supposed to be documenting.
Sometimes a mismatch is not a defect in the GIS. It is evidence that the historical record does not resolve cleanly. It may also be evidence of something deeper: that the contemporary parcel system and the histories carried by the whenua are not simply two versions of the same thing. A cadastral polygon is designed for a particular legal and administrative purpose. It should not be mistaken for a complete description of the relationships people have with that place.
That is where State of Unknown became useful. It gave us permission to stop treating ambiguity as a temporary technical inconvenience. A boundary could remain unresolved. A line could be indicative. Competing interpretations could be preserved. A note could explain that the historical plan and modern cadastre did not agree. The map did not have to pretend that one dataset had won.
This runs against the visual language of GIS. Maps want clean lines, databases want single values and project managers want completion. There is pressure at every stage to convert uncertainty into a definitive object. Yet in work involving whenua and historical claims, respecting uncertainty can itself be a form of respect for the people and histories represented. The responsible answer is sometimes that we do not know, or that the authority to resolve the question does not sit with the GIS practitioner.
Precision is not truth
Modern spatial systems are capable of extraordinary precision. GNSS positions can be recorded accurately, LiDAR can describe terrain in astonishing detail and satellite imagery can show individual objects. That technical precision is genuinely useful. The problem comes when people confuse precision with historical, cultural or relational certainty.
A polygon reconstructed from an uncertain plan does not become more certain because its vertices are stored to eight decimal places. A point can be positioned exactly where someone clicked while still representing a place whose actual extent is larger, whose meaning is relational, or whose location should not be made public. A place name can be perfectly centred on the map and still be wrong. A web map can be technically flawless and still communicate the wrong thing.
The distinction is simple: precision describes how exactly something has been represented. It does not tell you whether the representation is correct, appropriate or authoritative. In Māori GIS there is another question as well: does the representation preserve enough of the relationship to whenua, people and source knowledge to avoid turning something living into a disconnected spatial object?
This is why provenance matters. Where did the information come from? Who supplied it? What was the purpose of the source? Was it intended to be reused? Was the location exact, approximate or symbolic? Was the line reconstructed from one plan or several? Did people with local authority review it? Were there competing interpretations? The language of GIS calls this provenance and metadata. Those are useful concepts, but they are not substitutes for whakapapa. At best they help record part of the lineage of the information: where it came from, how it changed and whose authority needs to remain attached to it.
When those questions are missing, a GIS layer can become detached from the circumstances that gave it meaning. Once that happens, the geometry tends to survive longer than the explanation. Ten years later somebody opens a dataset, sees a polygon called ROHE_FINAL, and reasonably assumes that it is final because that is what the file says. File names have caused a remarkable amount of accidental certainty in the world.
Mapping with people
The strongest corrective was working directly with people. Claimant mapping was often iterative. Draft maps went out and comments returned. Researchers found more material. Claimants recognised errors. A field visit changed an interpretation. Someone would look at a draft and immediately know that a place had been positioned incorrectly or that a name was wrong.
One of the most useful sentences in mapping is simply, “No, that is not right.” It can prevent hours of sophisticated analysis from heading confidently in the wrong direction. Over time I came to see that not simply as quality assurance, but as part of a much more reciprocal way of working. The technical specialist was not the only person teaching. I might know the software, but another person knew the whenua, the history, the name or the reason the map mattered. Learning moved in more than one direction.
That is close to what I came to value in wānanga and later training as well: an element of ako rather than a one-way transfer from expert to learner. I do not use that word to claim a teaching philosophy that belongs to me. I use it because it describes something I repeatedly experienced. The strongest work happened when the mapper was willing to be corrected and when participants were able to shape the work rather than simply receive it.
This also changed the way I thought about requirements. In many technology projects, requirements are treated as something that can be gathered at the beginning and then frozen. Historical and claimant mapping rarely behaved like that. The act of mapping generated new questions. Seeing two sources together exposed contradictions. A map could prompt somebody to remember another place, locate another document or reconsider an earlier assumption. The mapping process was therefore part of the research and relationship, not merely a way of presenting research once it was complete.
Place names
Place names are a perfect example of how deceptively simple GIS can be. In a database a place name looks like text attached to a feature. In practice it can carry whakapapa, memory, language, history, colonial transcription, administrative replacement and local authority. The technical problem of displaying the correct label is often the easiest part.
During claimant work we encountered many corrections and additions to names. Later, place names became a recurring theme through Māori GIS conferences, Treaty settlement work, the New Zealand Geographic Board, the Gazetteer, Kā Huru Manu and iwi and hapū mapping projects. My own contribution in this area has mainly been technical and facilitative. I can configure the database, preserve macrons, make names searchable, compare historical and contemporary sources and help build tools that allow people to examine the information. None of that makes me the authority on what a place should be called.
This matters because a place name is not simply a convenient label stuck onto geography. Names can connect people to tīpuna, events, journeys and histories. A GIS may store the characters correctly, but it does not automatically carry the layers of meaning associated with them. The map can support the kōrero. It should not pretend to replace it.
The problem has become more important as mapping has moved online. An error once confined to a desktop project can now be copied through web services, applications and derivative datasets. Repetition gives bad information a strange form of authority. Once the same incorrect name appears on enough websites, people start assuming there must be a source behind it. Sometimes there is only a copy of a copy of a copy.
Permission and responsibility
Another lesson from claimant mapping was that having data and having the right to use it are not the same thing. GIS makes copying information almost effortless. A layer can be duplicated in seconds, moved into another project or published online with very little technical friction. The social and cultural responsibilities around that information do not move nearly as easily.
In claimant work, information could be supplied for a particular purpose and within a particular relationship. A map loaned for one project did not automatically become material for another. A culturally sensitive location mapped for an internal process did not become public data merely because it was now stored digitally. A layer available to the GIS team did not belong to the GIS team.
Later I encountered the language of Māori data sovereignty and rangatiratanga over Māori data, which gave a wider framework to many of these practical lessons. I would be cautious about retrofitting today's terminology onto every decision made twenty years ago, but the underlying principle was already visible: technical access does not create authority. The fact that I can copy a file tells me nothing about whether I should.
This is also where kaitiakitanga became relevant to how I thought about information, although not as a convenient Māori translation of “data stewardship”. The idea I took from the work was responsibility: holding information can create obligations about how it is protected, used, shared and passed on. The responsibility may extend beyond the immediate project and beyond the people currently sitting around the table.
That intergenerational dimension matters. A GIS project is often funded and managed over months. The whenua and relationships represented in it are not. A useful question is therefore not only whether the system works this year, but whether those who inherit the information will understand where it came from, what authority it carries and what limits were placed on its use.
Maps and authority
Maps have an unusual authority because people tend to believe them. Put a complicated statement in a report and a reader may challenge it. Put a line on a map and it begins to look official. Modern basemaps make this even stronger. The roads are accurate, the coastline is clean and the imagery is detailed, so whatever boundary is drawn on top inherits some of that credibility.
That can be dangerous when the overlaid information is indicative, contested or reconstructed. It can also be dangerous when the map quietly shifts authority. A line produced for one administrative purpose can begin to be treated as a definitive expression of mana whenua or cultural interest. A point supplied for internal use can become a public statement simply because it appears on a public map. The cartography can end up making decisions nobody explicitly agreed to make.
A map maker therefore has two responsibilities. The first is to display information clearly. The second is to avoid implying more certainty or authority than the information deserves. That may mean different symbology, notes, confidence fields, source descriptions, visible caveats or several interpretations rather than one. Sometimes it means refusing to use a neat line at all.
State of Unknown captured this better than any elaborate confidence model I have encountered since. It did not mean we had failed. It meant we had reached the edge of what the evidence, relationships and authority could support.
Digital certainty
The problem has not disappeared with better technology. In some ways it has become more important. Today's GIS can deliver national datasets instantly, stream imagery into a browser and combine hundreds of layers with very little effort. The data looks more authoritative because the systems presenting it are more polished. A layer can now be wrong at national scale, very quickly.
The same issue appears in artificial intelligence. An AI system can produce a smooth explanation of a place, boundary or historical event even when the evidence underneath is weak. The answer may sound much more confident than a careful researcher would be willing to sound. For Māori GIS the stakes can be higher than an ordinary technical error. A fabricated place-name explanation, invented boundary or confident summary of cultural information can become misleading very quickly.
State of Unknown is therefore surprisingly relevant to AI. The mature response to uncertainty is not always to generate another answer. Sometimes it is to stop, identify the gap and seek the person or source that can resolve it. Sometimes the question itself belongs with mana whenua rather than with the model, the database or the person operating it.
Better questions
Looking back, I began claimant mapping thinking the hard part was getting old information into a modern GIS. That was difficult, but it was not the most important challenge. The harder part was understanding what the information actually represented, what relationships sat behind it and what the map was entitled to claim.
Good mapping gradually became less about making every blank area disappear and more about asking better questions. Is this line legal, historical, indicative or interpretive? Is this name official, local, corrected or historic? Is this location precise, approximate or deliberately generalised? Is this information public, controlled or sensitive? Who has the authority to decide? What responsibility comes with holding it? What will the people who inherit this information need to know about how it was created?
Those questions have followed the work from historical claim maps through Māori GIS conferences, Indigenous Mapping Wānanga, hapū GIS and Ngā Poutama Matawhenua. The software has changed completely. The discipline has not.
GIS can draw almost anything now. That is not the same as knowing where the line belongs. Sometimes the best map is the one that admits the source is uncertain. Sometimes the best database preserves competing interpretations. Sometimes the most respectful publication decision is not to publish. Sometimes the strongest professional judgement is simply to say that the evidence stops here, or that the decision belongs to somebody else.
That is what State of Unknown came to mean for me. It is not where mapping failed. It is the point at which the map stopped pretending to know more than the whenua, the people and the evidence behind it.
For the broader history of this work, see From DVDs to AI: twenty years around Māori GIS, mapping and capability.