Above is a distribution of price differentials for the Dow Jones Industrial Average from the 1930s. The image was generated by one of my programs called Storm. I posted a few images from the same application in other blogs. If I recall correctly, the more volatile differentials (closer to the action) are at top; the more stable differentials (further from the action) are along the bottom. I describe this distribution scheme as a "hyperspatial image." The idea behind hyperspace according to my childhood exposure to several arcade games is to leap across distances instantly. It occurred to me one day to distribute a huge number of differential combinations across a thematic lattice, resulting in the rather surprising image above. The image is fairly persistent regardless of the type of data. On Storm, the image can be played like a view clip; so it is possible to see the migration of events across these mathematical rings. I believe that most people would say that the pattern hardly resembles the Dow Jones Industrial Average at all. Perhaps many would as I do suggest that it looks rather like the surface of water. The image has made me question the spatial nature of data and the issue of geography in data. In this blog, I will discuss how geography exists through conceptual delineations in data. I will then describe how a type of data-oriented geography might be used to map out the internal logic of certain phenomena.
From a coding standpoint, I find it difficult to distinguish geographic from any other type of data. I monitor the outcomes of my commuting decisions using a computer application. It is true that the routes contain geographic details; however on a coding level, there is no geography. Computer simulations routinely cause people to interact with geographic terrain that is purely fictional. Yet perhaps few would say that the terrain is any less geographic for existing only in a simulation. Moreover, the absence of "terrain" is not enough to prevent data from being described and handled in geographic terms. For instance, patrol reports from different police divisions might contain data characterized geographically but existing only in tables that can later be portrayed on a map. However, the data is no less geographic if it is never portrayed on a map. I suggest that the geography is not specifically about the terrain itself or even any data about the terrain; but rather, it is an aspect of how humans interact with the data. The fact that the data involves terrain is not necessarily important. There doesn't have to be any geography (in a material sense) in order for data to be interpreted geographically (in a logical sense); this distinction raises the question of what it means for something to be "geographic." I offer an interpretation where geography is mostly about the organization of data.
Consider the problem of schoolyard bullying. For to me, this is an issue that is geographically or at least spatially relevant. The same can be said in regards to incidents of sexual assault on campus. However, data regarding these types of incidents seem suspiciously deprived of any geographic context. Geography is arguably not an issue in relation to bullying given that online bullying is so common; and of course the online world is notable for its lack of anything resembling geography. The assertion is perhaps harder to make in relation to sexual assault given how easy it is to find secluded areas of poor lighting where predators might hide; still, I believe that many would say, sexual assault can likewise occur online without any apparent geography or spatiality. Rather than argue whether or not online events can be characterized geographically given lack of physicality, instead, I challenge the premise that geography must always be tangible or material. People certainly encounter delineations online. Homogeneity is difficult to find. Online, people are separated by language, their interests, social objectives, philosophical leanings, and their professions. I find it difficult to accept that the absence of a tangible world renders such delineations non-geographic.
I was born on a complex arrangement of tropical islands containing three distinctive features: land, water, and volcanoes. I recall an old photograph of me near a small enclosure I believe designed to smoke fish; the water stood just a short distance away. I don’t remember posing for the photograph. But I recall the volcano at the other side of the water. Many might describe geography as my placement in relation to the land. But consider instead the placement of the land in relation to my perception of myself. My recollection of the scene is indeed geographic; however, the images and feelings are entirely from "data" - if memories can be described in this manner. This is not just any data like the length of car: it is embodied data conveyed though lived experience. Perceptual context can be connected to massive amounts of data. The big data of the discursive is the biggest data of all; there are no boundaries either to discourse or self-placement. The geography of data as I envision it is about self-placement in the expanse of data; more specifically, it is about the relevance of how data is arranged or structured in relation to the person or organization.
Self-placement in data is an important concept. I have a couple of illustrations to help with the conceptualization. The first image is called "ME Attribution," which stands for mind-environment attribution. When we have a piece of data, it is often worthwhile to ascertain whether it is connected with the mind (M), environment (E), or some aspect of both (ME). The more environmental the data, the easier it is connect the data to physical landmarks. For instance, a phone-booth, lamp-post, and birch tree are objects that are usually physically connected to the material world; it is therefore possible to create a map showing the geographic distribution. My perceptions of the same items can be found my mind. The geography that these items occupy exists within me instead of the material reality. In fact, I assert that such things would have no place within me were it not for the geography already existing inside.
The second illustration is called "BME Attribution" for body-mind-environment attribution. In my ME example, I was describing the placement of data at polar opposites: the psychological versus the material. Data can often be connected to the body (B) or the physical. Given that the mind and body are so close together, and since one occupies the other, at times it might not be possible to make clear separations. Injury to the body frequently leads to mental distress. Mental strain can erode the body. Since the body occupies an environment and the mind lives in the body, it makes sense for the mind to feel stress upon the destruction of a habitat for which its body is optimized to occupy. The placement of a person in unfamiliar physical environments such as a busy office or noisy factory might contribute to stress. Similarly, I suggest that self-placement in a confusing and hostile data environment can also cause stress. (Self-placement in this sense does not mean that the person takes occupancy; but rather he or she finds the self placed in the environment.) The geography of data serves not just to provide access to data but distance from it; make it possible to place data in front and behind other data; create meandering routes and straight paths. Body, mind, and environment represent reference points to help us understand the placement of the person in relation to the data.
Sometimes a problem might be attributed to the body while it is actually environmental in nature. Consider for instance the concept of a revenue stream: increases in revenue tend to be associated with growth and expansion. However, such revenue streams might not reflect prudent decision-making specifically. The financial data could merely indicate that there is market growth: a company that operates during good times would tend to have good numbers. Revenue data offering metrics of the market can instead be interpreted as metrics of the market participants. Conversely, data pointing to poor performance might reflect the market more than the participants that occupy it. Consider for example the plight of a technician that services air-conditioning systems. I sometimes give HVAC examples in my blogs perhaps due to some vocational training as an HVAC technician. The HVAC technician scorns warm winters and cool summers. It’s not due to lack of skill that prevents him or her from putting food on the table but rather moderate temperatures.
I believe that studying the geography of data has some useful applications. It is possible to "map out" the geography of non-spatial environments. The mapping of non-spatial spaces can be achieved through a process that I call "tracing." When events are grouped together as a class, interesting insights can be gained such as the types of future events that might be thrown in order to detect particular aspects phenomena. I introduced the ME and BME attribution models not just to give an explanation of placement but also how events to detect phenomena can be distributed. For instance, taking a pill in order to become healthier is a concept entirely focused on the body. After some period of use, a person might discover that food supplements have few positive or negative impacts on health. However, exercise could prove to be much more relevant than food supplements. Exercise is "less" body-oriented than taking pills since it requires space outside the body to perform: e.g. jogging, swimming, and even weight lifting. Classes can be filled with events to delineate divisions within exercise; and perhaps at some point, it would become apparent how some forms of exercise are better than others. So there can be a perpetual process of both of expansion and precision. I’m uncertain if a term already exists to describe this pattern of discovery. For the sake of argument, consider "dendritic epistemology." Once again I am thwarted by a college dictionary short on scientific terms. Anyways, the roots of a tree expand not through random firings but rather progressive extension. Perhaps a biologist or botanist can pass along the term to me.
So the geography of data is about self-placement within data, which means finding the relevance of events in relation to the person or organization. I actually have a more formal term for this type of geography. I call it "Transpositional Geography." This geography can be "reconstructive" as in the case of computer simulations. I know that geography can also be "deconstructive," and I don’t dispute how I seem to be describing an approach that could be used deconstructively. As far as data science goes, I think the market is in reconstruction - helping organizations "realign." A spatial scene can be reconstructed from data assuming that the data is suitably configured; for me, the same approach can be applied in relation to scenes that are non-spatial. For me, Transpositional Geography is focused on the data as an area of study in itself particularly for the purpose of mapping non-spatial events. This should be a topic of some interest to data scientists that regard the internet, cloud systems, and networks as important sources of data. I would argue that the evolution of the internet has been about finding and placing self in the limitless expanse. Many of us experience a sense of belonging to things in the material world; in the absence of physicality, for instance on a website for an online retailer, the idea of belonging should be a concern given how easily people can come and go as they please.
Data does not have to be interpreted geographically - that is to say, transpositionally – or, dare I say transpositively. In fact, some data might leave little room for self; but then nobody would care about it. I remember during Sunday service one day being fascinated by the layout, signs, and symbols. Although this was a service involving physical bodies in a place of worship, the scene and setting seemed intended to trigger much deeper connections. The non-spatial component of the service could take me places beyond the confines of physicality. The service leader helped people organize the placement of things inside themselves. He brought some objects closer while taking other objects farther away; put some items in front and others at the back; he forced people to retrace their steps. So the ceremonial discourse seemed highly geographic in nature; everything seemed to matter much more.
Now, I was inside something like a mechanical warrior on an alien planet. This was a simulation of course. It had nothing to do with Sunday service by the way. There was enormous landscape all around me and big sky high above. Mechanical warrior planes were flying overhead and shooting. They were on my radar. I was shooting back although not too well. It became apparent to me that space was not behaving properly on this planet. I could reach destinations rather quickly. Space appeared a bit warped; or my ability to move through it was accelerated. The scenery was present to add realism; but since my main objective was to engage in combat rather than move through scenery, the scenery allowed for abnormally rapid passage. I therefore argue that the existence of spatiality in this simulation was merely to support transposition. I further contend that when people surround themselves with facts and figures, they do so not merely for utility but also transposition.
I would expect Transpositional Geography to be useful in relation to investments. There is often a need to map out the internal logic of the market. In the day-to-day price fluctuations of a particular stock, information about the underlying company tends to be released relatively infrequently. It is therefore reasonable to assume that the dynamics outside the body (B-company) that can be responsible for sifts in valuation. Consequently, spending a great deal of time and effort on the internal events of the company, for instance focusing on the valuation of assets and revenue streams, is problematic given the influence of the environment. At the same time, an entirely technical approach making use of market trading data seems insulated from information about reality. I would suggest that a more geographic strategy considers the issue of market self-placement: this is the setting in which information about the company comes to light and also details about competing needs for capital. I find it simplistic to interpret pricing strictly in relation to the "investment" and never the "investor" (MBE-market). Sometimes investors sell not because of the investment but rather the need for funds, perceived risks to themselves, and lack of connection to "the dream." Nonetheless, part of the joy of a geographic approach is to survey the internal logic of something in a literal sense rather than having somebody like me defining it. "They are selling because they think they are made of blue cheese." Ah, that assertion seems inconsistent with the non-spatial delineation of relevance.
Another useful application that comes to mind would be in situations of sick-building syndrome. In some of these cases, employees might be blamed for poor performance and lack or attention to detail when in fact the ventilation might be poor, or parts of the building are toxic. I encountered a similar problem in my own research around the time when computers were becoming prevalent in office environments. Almost all metrics of performance were criteria-based, as they still are. In such a paradigm, it would seem that improving performance would simply be a matter of raising the bar and adjusting behaviours. However, I found some evidence showing that about half of insurance disability claims were related to musculoskeletal and circulatory problems; these indicate BE, BME, ME, and E factors. Therefore, one would think that attitudinal metrics might be less relevant than environmental considerations. I am not dismissing the practice of providing psychological counselling and support except that it might sometimes be completely unrelated to the underlying problem and therefore unlikely to lead to improvements in performance. Stated differently, in some cases it might be a waste of time and money. Certainly, providing counselling but not protective equipment is rather negligent in relation to asbestos and exposure to radiation. It is important to delineate data in a transpositional manner rather than expect it to satisfy professional preconceptions.
Transpositional Geography can help us understand the relevance of things to an organization. We can study the data present in a department for indications of perceived value. It is also possible to assess the relevance of the data to actual contexts for instance using a tracing technique. I feel it is possible to "explore" data through tracing in order to thoughtfully arrive at relevant events and contexts. This is not to say that the data collected would necessarily be "small." However, finding worthwhile data would be a much more deliberate process with coherent objectives compared to something like data-mining. Transposition allows for the mapping of "local" perspectives. It is not an instrument that imposes meaning over the data; but rather it is a method of exploration. By local, I mean that transposition supports local sensitivity, the decentralization of information, and processes involved in the articulation of outcomes and consequences. In my blogs, I discussed the topic of "data embodiment." The geography of data is about self-placement and the organization of data around the self, which as I said might have nothing to do with the material world. The self can exist online and in the cloud. While people still migrate across oceans and over landscapes, these days they move effortlessly through seas of electronic data; venture over convective gales of information; not lost but aware of their own placement and the things important to them.