When my partner calls me on my cell phone, I am often riding my bike and the question she first asks is: where are you? The possible answers to this question illustrate the many ways that place can be constructed. The most obvious answer might be that I am riding along street x, or at intersection y. However, she might be calling when I am riding into a headwind, during a snowstorm, when the temperature is â€“10, when I am going over a log on an off-road trail or when I am riding in the midst of 30-40 other bikes at fairly high speeds. Now, of course, I would probably not be answering my phone when in these situations. A more complete, meaning rich answer to the question: Where are you? might not even include the actual location but could include details about the context. An additional response might not even specify where I am but where I am going. I am going to York, I am going downtown, or simply I am on my bike.
I started gathering data for oh, those everyday spaces, in January 2002 and stopped in March 2003. During that period, I gathered about 30,000 images, 80,000 GPS (Global Positioning System) locations and hourly weather data from 3 local stations. In their raw form, the images alone were 10 gigabytes. Of course, I needed a way to store, manipulate and access this data in very flexible ways. So I chose an open-source web-based database, MySQL and PHP. But the database is not just an innocent conduit for storing data, it is a computer and cultural form that encompasses implicit ways of structuring and conceptualizing data and hence, influences our ideas of what can be achieved with databases.
What is a database?
The most straightforward definition of a database is simply: a collection of structured data. In The Language of New Media, Lev Manovich discusses database and narrative as cultural forms and explores the relationships that exist between these forms. Manovich begins his argument by stating that “As a cultural form, the database represents the world as a list of items, and it refuses to order this list. In contrast, a narrative creates a cause-and-effect trajectory of seemingly unordered items (events)” (p. 225. Manovich, 2002). He goes on to observe that new media objects are often database-like collections of individual items, with each item having the same significance in relation to the whole. New media works are easily built so that all of the content is placed in a database structure. Within that structure, narrative becomes just one way of accessing data among many others. After an extensive discussion of database and narrative, Manovich concludes that database and narrative are “two competing imaginations, two basic creative impulses, two essential responses to the world”.
What are the implications for my work? First, while a database is a list of items, despite what Manovich says, in practice one must create a default ordering forced by the necessity of assigning a unique primary key. Given the nature of my dataset, there is an obvious way of ordering the data: time, since data were gathered periodically. Simply playing back my experience using this default ordering is not terribly interesting because it does not allow one to engage with the complex relationships between the data. The dataset does have an inherent spatial quality since each data point has a location in space. Spatial data is much more difficult to work with in a database because it uses two dimensions. How do you order spatial data? If one imagines a database of first and last names, ordering is straightforward: sort on last name then sort on first name. With spatial data, although it does have two components (east and north) sorting by east then sorting by north does not yield useful results. Because of the kinds of data in the data set, I needed to find ways to think about the dataset that would allow me to explore it to its fullest.
Sorting or ordering the events in the database becomes the challenge. Interesting ways of ordering events allows one to present the material in the database in ways that reconstruct the experiences by creating relationships that were not initially obvious.
The oh, those everyday spaces database is comprised of a number of fields. These fields are of two types: raw data and derived data. Raw data has not been manipulated and includes: image reference (pointing to an image file), imagedatetime, position components (zone1, zone2, easting, and northing), temperature, condition, visibility, humidity, dewpoint, wind direction, wind speed, and wind gust. Derived data is calculated from raw data and includes: speed (distance/time), direction, avgdirection, gpspointdiff. A random sampling of the dataspace is displayed on the right.
Ordering by speed or temperature for example, creates relationships through grouping experiences based on those parameters. For instance, what are the various situations where I would be traveling 10 km/hr, 25 km/hr or 50 km/hr? Things get quite a bit more complicated when one wants to ask and answer questions like: Given that you are traveling along a particular road, what is the next datapoint in space? What I really needed was a way of thinking about the data that would help me develop strategies for working with these questions.
Conceptualizing the dataset
The most useful paradigm I found was the idea of an n-dimensional space. This paradigm involves thinking of each field in a database as one dimension. The n-dimensional space is most easily understood through building the number of dimensions from 1. In a 1-dimensional space we can move forward and backward. In a 2-dimensional space left and right is added. In a 3-dimensional space, up and down is added. The next step is to move away from space as a metaphor of dimension to a generalized idea of dimension. A dimension is now simply movement through the data on one field be it temperature, easting or time for example. Using this paradigm does not change the database itself but, more significantly my way of thinking about it changes and I can construct queries based on this idea.
Variations/Variantes takes the idea of navigating through a multi-dimensional dataspace, as I call it, and provides an interface to 4 specific dimensions. The idea for the piece comes from a tool in Adobe Photoshop that allows one to navigate through an image’s colour space.
|Photoshop Variations Window|
As one clicks on an image surrounding the Current Pick, one is traveling through an image’s colour space. Images around the current pick change to reflect the current location in the colour space. Just as Photoshop’s variations tool navigates through an image’s colour space based on colour and brightness, Variations/Variantes provides an interface to navigate through the oh, those everyday spaces dataspace in ways that focus on the unique qualities of the collection. Qualities like wind, temperature, and traveling speed are factors that affect a cyclist’s experience of any place. As one clicks on an image surrounding the current pick one travels around oh, those everyday spaces exploring relationships based on the dimension being investigated. As noted earlier, movements forward and backward are not really along one dimension but are derived from the position and direction of a datapoint. The design of the interface serves to emphasize that it is the act of navigating, traveling through the data that is key, with the image presented after clicking OK being an anticlimax.
10 Second OTES
10 Second OTES is a series of nine short image sequences that explore a number of locations in Toronto. After creating Variations/Variantes, I became interested in the way images could be organized around specific locations. Each of the sequences in this work explores the rhythm of movement through space. In one way, rhythms are created through the frame rate of image playback. The relationship between distance travelled between images and the number of images presented over time evokes various levels of intensity relating to the experience of cycling. The visual rhythm of images themselves adds a layer of interest expressing conditions including sloppy wet snow, bright days, and dark nights. I particularly enjoyed how each different sequence had a unique way of combining day and night images. The name of each sequence is based on its' GPS location. My work on 10 Second OTES encouraged me to create work elaborating on these ideas.
During the time I was collecting data, places that I traveled repeatedly were much better represented. My daily commuting route was particularly interesting because it produced a large number of data points along specific routes over the course of 15 months. Going Home focuses in on part of my commuting routine. Aside from the nice title, I picked the route home as opposed to the route to York because it provided a more varied set of images. I tend to go to work when it is light but when I go home it could be light, sunset or dark. To produce this work I set up a database query that would return a series of images that were in front of any given position. Doing this repeatedly from a position along my commuting route gave me a large number of data points ordered by position along the route. Whereas Variations/Variantes was a visual piece, Going Home has a sound track based on data. To create the sound track, I used the speed I was traveling when each image was taken to create a number of layered sounds. Each one of the sound layers is an expression of speed over varying time periods. Going Home has two versions: a video and a video installation. The interface to the installation tracks movement in the installation space; the more movement, the slower the work progresses. This creates an inverse relationship between the participant and the space. Passivity causes intensity and activity promotes contemplation of detail.
Conclusion and next steps
In the time I have been working on this project, I have been able to find connections to a number of current themes. Two of those themes are: image metadata and inverse surveillance.
While working on this project, I have tried to think of all data in the dataspace as relatively equal in value. Image metadata is data that accompanies an image to describe it. Recent writings on image metadata (metadata.net) have reminded me that, while extremely important decisions need to be made about how we describe images, my approach is fundamentally different. I have no metadata for my images. I could even think of the image itself as metadata for the data. I do not want to be forced to decide that any one part of the data is more important than another since making such a decision would limit the scope of my exploration.
Inverse surveillance or sousveillance is the personal capture of everyday life. Probably the most well-known souveillance practitioner and advocate is Steve Mann. My goal is to help define artistic agendas to help prevent development of this area from being solely consumer-based. To that end I am working to develop a more integrated and diverse data gathering wearable in the near future so that I can further investigate specifically artistic approaches.
Manovich, Lev. The Language of New Media. Cambridge, Massachusetts: MIT Press, 2002.