Summary

In Study 7/0 we leverage Global Positioning System (GPS) errors as a source of generative emergence to indicate the tensions between navigation technologies and their users’ individual directions and interests. We gathered the error data by running a TrackLog function on a fixed GPS receiver for seven days nonstop. Instead of a single location it should have ideally registered, our TrackLog captured 8,438 distinct positions, indicating motion on a 34.7 km trajectory that covered an area of 2.1 km² with an average speed of 200 m/h. This is a combined consequence of the GPS device’s operation under changing weather conditions and the GPS infrastructure’s intrinsic limitations.

Using the longitude and latitude of the collected waypoints and proportionally adjusting their original timestamp deltas to reach about five minutes of total running time, we animated a red dot along the horizontal projection of this wandering path. We contextualized the visualization with a synchronous display of all TrackLogged spatiotemporal values, including altitude, heading, distance, date/time, speed, etc. To generate sound, we treated the virtual, data-implied movement as a traversal of a listening device in real space with the ensuing acoustic phenomena: Doppler effects and volume modifications. The resulting work is an installation comprising a large-scale video projection and a two-channel audio.

Artists: Dejan Grba & Philippe Kocher

Concept

Study 7/0 belongs to a series of generative artworks that investigate emergence in complex systems, where relatively simple configurations can produce surprising phenomena (Grba 2015, Boyé et al. 2019). The series began with Study (2010), which focused on cognitive mapping as a configuration of individual, non-linear, and discontinuous spatiotemporal experiences and explored its implications in digital culture. It combined a selection of geopaths of IP routes toward the websites we visited during one week with the TrackLog path that a GPS receiver on our desk recorded in that period (Grba 2010a).

As a wayfinding framework, the GPS integrates technical knowledge and infrastructure with a layered set of often-conflicting interactions, choices, and compromises that constitute contemporary life. While its common applications rely on accuracy, in Study 7/0 we focus on imprecision as an inverse, but equally revealing index of its usefulness. The interplay of shape, space, and signs in our three-dimensional rendering of a four-dimensional entity indicates the necessary yet problematic information collapse in wayfinding and mapping representations. GPS tools calculate routes as tradeoffs between environmental constraints and desired accessibility, so their slipups incur practical damage and, in a broader sense, also highlight the mismatch between our diverse personal intents (idiosyncratic identities) and our technologies’ pragmatic but often coercive functional requirements. Our subjective negotiation with technical systems’ instrumental constraints and adaptive correction of their shortcomings also invokes the notion that politics, before it becomes public, is always a personal, intimate matter and should be primarily resolved on that level (Grba 2010b).

By prioritizing the ideational aspects of systemic GPS errors, our expressive approach in this series contrasts glitch art practices, which predominantly foreground the aesthetic effects of actively inflicted technical malfunctions (see Barker 2012).

Dataset

To gather the error dataset, we secured a Garmin GPSmap 60Cx GPS receiver to a desk, powered it from the outlet, selected the TrackLog function, and let it run for 7 days, 7 hours, 16 minutes, and 11 seconds (from 7 July 2010 4:46:36 PM to 15 July 12:02:47 AM). Garmin GPSmap 60Cx is a handheld GPS navigation device, well-regarded for its high sensitivity, accuracy, functionality, ruggedness, and reliability. TrackLog function starts as soon as the receiver gets a satellite location fix and saves the time, location, elevation, distance, and speed data about waypoints, which it creates according to a selected sampling method: by time period, frequency, or distance.

If the sampling method is time or frequency, the ideal TrackLog record on a fixed GPS receiver is a series of overlapping timestamped waypoints with identical location and elevation, zero distance, and speed. If the sampling method is distance, the ideal record is a single timestamped waypoint (location and elevation), with zero distance and speed. However, our TrackLog registered 8,438 different positions, indicating movement on a 34.7 km long path that covered an area of 2.1 km2 with an average speed of 200 m/h and a maximum speed of 17.9 km/h (figures 1 and 2).

1. TrackLog path.

2. Elevation data.

This is a combined consequence of the GPS device’s operation under changing weather conditions and the GPS infrastructure’s intrinsic limitations (see Thin et al. 2016).

Animation

We animated a red dot positional marker along the TrackLog path’s horizontal projection, defined by waypoints’ longitude and latitude (figures 3 and 4, high-res PDF ), but proportionally adjusted their timestamp deltas to get the total effective running time of 281.233 seconds (4 minutes and 41 seconds at 30 fps) from the originally recorded 630,971 seconds.

3. Horizontal proj.

4. Horizontal projection, detail.

As it moves, the positional marker leaves a black trail on the last 2.25 percent (equivalent to 780.75 meters) of the trajectory it traversed. Initially rendered as smooth curves, its directional/angular changes gradually morph toward the jaggedness of the actual error path.

5. Animation frame.

6. Animation’s left section, detail.

The animation juxtaposes the positional marker’s isolated trail on the left with its overlap on a gray-colored buildup of the whole path on the right. We contextualized the animation with a synchronous display of all TrackLogged values, including altitude, heading, distance, date/time, speed, etc. These visual elements reveal the intricate dynamics of error-generated virtual motion (figures 5 and 6).

Sound

To generate sound, we treated the data-implied movement as a traversal of a listening device in real space. This allowed us to exploit the ensuing acoustic phenomena such as Doppler effects and volume modifications. The Doppler effect is the change in the wave frequency due to a change in the relative distance between the source and the listener. As the distance decreases, sound waves take less time to reach the listener and get condensed, increasing the frequency and causing the perception of a higher pitch. As the distance increases, sound waves are more spread out, the frequency is lowered, and a lower pitch is perceived. Additionally, the increasing distance between the sound source and listener decreases the overall sound volume, while the energy loss of sound as it propagates through air attenuates high frequencies.

We created a metaphorical analogy to these acoustic phenomena using delay lines whose variable delay times lead to shifts in the perceived pitch (the Doppler effect) and low-pass filters whose variable cutoff frequency recreates the air absorption effect. However, we did not intend to establish authenticity by building a proper physical model. All these devices were applied freely as individual sound design components and scaled according to our considerations for minimal sound aesthetics. We searched the TrackLog dataset for the waypoints’ maximum positional deviation to define a square with the GPS receiver’s actual/true geospatial location in its center. Four virtual sound sources (S1–S4) were placed in each corner of this square, and the horizontal waypoint position data controlled the movement of a virtual microphone (M). The resulting distances (D1–D4) form the sound processing parameters (figure 7). We normalized the microphone coordinates to the range [-1,1] relative to the dimensions of the max-deviation square and thus calculated the distances between the microphone and sound sources. Each recorded signal was subsequently attenuated, filtered, and delayed based on the respective distance (figure 8).

7. Sound signal sources.

8. Signal flow.

The sound sources emit band-limited pulse trains at 3, 5, 7, and 11 Hz. This superimposition of prime number frequencies yields a compact rhythmical texture. In time, the four pulse trains become slightly faster, creating a slow but steady increase in density. This compensates for listeners' gradual habituation to the sound quality and corresponds with the visual accumulation and increasing jaggedness of the animated path. We made final enhancements by applying a high-pass filter to render the sound less booming, distributing sound signals among the two channels as if the virtual microphone were a stereo microphone, and adding a subtle reverberation.

Notes

Detailed project description PDF

Barker, Tim. 2012. “Aesthetics of the Error: Media Art, the Machine, the Unforeseen, and the Errant.” in Nunes, Mark, ed. Error: Glitch, Noise and Jam in New Media Cultures. New York and London: Bloomsbury Academic: 42-58.

Boyé, Paul, Dejan Grba, Kristy H. A. Kang, Melentie Pandilovski and Vladimir Todorović. 2019. “Machine Flaws in Generative Art.” In Juyong Park, Juhan Nam and Jin Wan Park, eds. Lux Aeterna, ISEA2019 25th International Symposium on Electronic Art proceedings. Gwangju, Korea: 713-716.

Grba, Dejan. 2010a. “I Know What I Drew Last Summer.” Rhizome of the City exhibition catalogue. Belgrade: Museum of Science and Technology.

Grba, Dejan. 2010b. “Study.” Dejan Grba’s website. https://dejangrba.org/art-projects/en/2010-study/index.php.

Grba, Dejan. 2015. “Get Lucky: Cognitive Aspects of Generative Art.” In Celestino Sodu, ed. XIX Generative Art Conference Proceedings. Venice: Fondazione Bevilacqua La Masa: 200-213. https://dejangrba.org/lectures/en/2015-get-lucky.php.

Thin, Li Nyen, Lau Ying Ting, Nor Adila Husna and Mohd Heikal Husin. 2016. “GPS Systems Literature: Inaccuracy Factors and Effective Solutions.” International Journal of Computer Networks & Communications (IJCNC). 8, (2), March: 123-131.