Transitions
On a map, lines can connect or divide. But the most interesting areas to explore are between those lines, in the transition.
I was lucky enough to join the conversation in the Geospatial Innovation LinkedIn group this month. With Linda, Tim, Bonny and special guest Bill (Unfortunately, Tee could not make it) we talked about maps. For most speakers, this turned into sharing our origin stories: how and why we found ourselves in mapping. But with these stories, challenging reflections emerged. One of the interesting themes I saw was in transitions - the spaces between.
Typically, in mapping, we refer to the two data archetypes of "raster" and "vector." As a result, we will often see landscapes as something or not, this or that. A polygon has a value, and its edge will define where that value changes. A pixel holds a value, and its neighbours might hold the same or a different value. In raster-land, we only hold a surface of a single variable and typically visualize those through one of three colour guns. Of course, there is a history of a middle ground in the form of the “raster attribute table." And, yes, an image can hold multiple bands, even hundreds, in the case of hyperspectral sensors. But for the most part, our typical geographic information Systems (GIS) and geospatial datatypes try to reduce our landscapes to simple models. This is reasonable, as a map is a model, an abstraction, of our more complicated world. That has always been the basic idea.
The discussion soon moved to the defining features of a map. While the dictionary has one particular description, when I think about the basic nature of a map, I think about a graph. If a graph is viewed within a shared geographic context, then it becomes a map of sorts.
Because, remember, maps are for people, so the shared context is important. If we designed geographic information systems for machines, there would be no map whatsoever.
So, this graph could be a sketch on a napkin, a line in the sand, or a precise (and perhaps even accurate) cadastral representation. In each case, the graph users must have a shared geographic reference, converting the coordinate system used into a map. The information held within that map is layered on the agreed reference system. Perhaps you have added navigational points of interest to the napkin map or a notable landscape feature to the sand map.
An observation I have often made of maps is how limited they can be, especially in representing natural features. Again, the limitations can provide a powerful simplicity for singular purposes. But in an age of practically infinite compute, should we continue to accept limiting data types?
A quick side note on culture
Interestingly, when considering simplicities, we could also observe that human philosophies tend towards binaryisms and extremity. We like to classify and differentiate things. I am this, you are that. I am even more this than Bob, who is only a little bit this. People follow colours. Generations have letters, yet are obviously measured against a continuous scale. We are all different types of professionals and different types of people; we like to classify and then reclassify.
Landscapes are not binary
But, natural landscapes are more likely to transition constantly. Let me try to explain. Countries have borders: I am in a country or not. granted, those borders can be disputed, but that dispute is not a shade of grey between the disputees. Both countries are laying a claim to that piece of land. I am not 50% in one country and 50% in another. So, countries are typically binary. Likewise, I could be in a municipality or not. I am in a polygon, or not. Polygons have discrete, almost infinitely precise edges. Lines that divide or connect.
What about the transitions?
We could argue that the Euro-centric, reductionist view of our planet has brought us to this interesting but ultimately limited place. Having worked in British Columbia (BC) for some time now, I have been humbled working alongside several First Nations on land management issues. However, I would encounter other “GIS” professionals who might smirk when the observation was made that more than 110% of BC's land base was being claimed in the treaty process. How could that be possible? Is this just greed, territorialism, or expansionism?
No. This illustrates a misalignment of concepts: two separately developed philosophies colliding. The euro-centric view is to put a line on a map: to own. To be this or that. The First Nations view was more about seasonal presence on a piece of land, so it’s perfectly reasonable that two or more Nations might use the same piece of land, each without owning it, simply moving through the landscape because hunting was good for a time. Ownership was unnecessary. Use was key. So when, during what is called the treaty process, the Nations were asked how their "territory" could be represented within the limited euro-centric data models discussed above, they naturally drew a line around the entire area they would move and hunt. Resulting in large overlapping areas. We can see here colliding philosophies driving division between cultures, and sub-cultures. Nations would be forced to argue over assets that would traditionally be shared. Indeed, they would never have been seen as assets but as gifts.
There are numerous other examples of our community drawing lines where their presence is misleading. But clearly, in terms of a typical geospatial representation, the traditional territory should be represented by a surface of activity rather than a line. But if we consider the number of Nations, then quickly our data models start to fail.
With this simple extrapolation of philosophy, our visualization tools start to demonstrate their limitations. In fact, in many ways, our visualization paradigms are failing across the board. Our present tools are built with a particular reductionist philosophy in mind. Perhaps we need to rethink them entirely for a more holistic and, in fact, truer representation of our landscape.
In ecology and forestry, we draw lines around similar-looking stands of vegetation. But those lines don't exist; nature is in almost constant transition. Clearly, reductionism and our tooling will start to fail as we enter a time of broader AI adoption, and we start to ingest larger datasets to begin to address the complexities of climate change.
Our future data models need to understand and model multiple transitions simultaneously. We must interpret environmental, climactic, and human nuance at a planetary scale.
From this article, you can pull three ideas:
Our most commonly used geographic representations are flawed, and better technical representations are necessary to solve bigger problems.
As we move from binaryism to a more sophisticated, nuanced representation of our landscapes, better outcomes can be created.
Perhaps there are other parts of our lives where a more nuanced view would provide a better perspective on events.
Finally, a quick thanks to Tee Barr for editing support.