Here we see the probe being calibrated before setting out into the city.

This probe was built in one day, and images were tested on the same night. It shifted our working pace from long iterations up to a very fast overnight turnaround of new experiments. This was an important and much-needed shift in our exploratory process. With a probe that was light and fast, it was possible for us to go out and explore the city.

In repeating walks many times along the same routes, we discovered that we couldn’t get repeatable readings. Sometimes accuracy would go down next to a building; sometimes it would go up. There were many reasons for this, but the key reason is that GPS accuracy varies just as much (perhaps more) over time as it does in relation to space.

When read in sequence, our new visual mappings were now temporal as well as spatial. They showed us how GPS accuracy changed within an urban environment over time. With the long exposure times, our light painting technique was not only conflating time and space, but was also exaggerating temporal errors and latency in the GPS system. We realized that we had to find new techniques to more effectively show changes over time.

Timelapse

What we discovered with our more nimble orb probe is that GPS in the city is temporally unstable; as satellites pass overhead and get occluded by the urban environment, the uncertainty changes significantly. This required us to reconsider GPS as a temporal phenomena. We needed to find ways to measure GPS accuracy from various points and to look for patterns over time.

Some initial questions arose: what are the timescales over which significant changes occur? Twenty minutes, one hour, five hours? We decided to build new software tools for mapping GPS uncertainty over time, and in this next gallery we look at how these tools evolved.

In these graphs of a stationary GPS receiver, we saw patterns that related to the movement of satellites overhead. We recorded many of these graphs in different locations, and could observe how the various physical environments affected the patterns in accuracy shifts.

To be able to show these patterns in context, we wanted to begin to use photography and filmmaking techniques. We knew that timelapse was a technique to compress phenomena that changes or moves over time. It’s popular in nature films, astronomy or as a narrative device in TV and cinema to show that time is passing. But, it has also been used for more experimental purposes, as in Michael Marantz's (2012) NYC Dark.

As our aim is to communicate about GPS, not just graph it, we wrote more software to be able to see what these changes would look like if they were turned into colored orbs and shown over time (from black to white).

We created some experiments with timelapses in code, using Processing software, shown above, where we turned accuracy values into the brightness of a graphic circle. This showed a really promising visualisation of the granularity of GPS at a particular point. It felt like we were finally seeing some of the material of GPS signals.

Inspired by this, we built a simple modification of our orb probe, a stationary lamp with a GPS receiver that would change color or brightness depending on GPS accuracy.

In this short sequence, we can clearly see how accuracy fluctuates over time. From our notebook:

We have made something that is beautiful, understandable and exciting. (Notebook, 2012)

Beyond this interesting timelapse, we also found that the lamp itself was a fascinating object to observe as a physical object. Sitting beside it, over the course of a few days, as it was responding in real time to GPS signals, we started to gain a nuanced understanding of how GPS was inhabiting different spaces. The lamp worked as a visualization of the specific accuracy pattern created by the environment in which it was placed.

We had finally found a method that reflected a material quality of GPS signals. We could also give a one-line explanation of what you see in the film:

This lamp is controlled by satellites

Spatial and Temporal

From our initial ideas of heatmaps, to explorations of spatial diagrams through a glowing orb, our development reached a point where we had to admit we didn’t yet have any visualizations that were communicative. Heatmaps were hard to read, the qualities they showed mixed the spatial and the temporal uncertainty in ways that muddled any explanation. They accurately reflected the odd phenomena of GPS, but not in a meaningful, communicative way.

So through iteration of our approach, we arrived at a more communicative way of showing the change in accuracy over time. This approach was measuring the temporal qualities of GPS accuracy, through a set of discrete probes and single points of light that could be photographed over time as timelapses. We had discovered a foundation for our communicative approach to GPS as an urban phenomena.

On Communicative Forms

Our first iterations of our GPS visualizations showed that we needed to rethink our communicative form. In this section, we have outlined processes of data-gathering and probing. Our experimentation led to a shift in focus from the spatial to the temporal.

Earlier we have argued that the spatial qualities of WiFi can be externalized through similar experimentation with rigs and light painting (Martinussen, 2012). In contrast, in this chapter we have shown that it is the combination of GPS’ temporal and spatial qualities that need to be given closer attention. This required a different set of design responses, and was an important turn in working to make visible the material and dynamic qualities of GPS. As Latour (2013) has repeatedly stated, we need to consistently challenge the surface assemblies of technology and expose their technological workings.