Final Project Documentation: “Global IceCubes”

“Global IceCubes” in its demo/final project form, was an installation where the user could enjoy a relaxing drink of scotch on the rocks…however, after the first sip is taken, the visitor watches their ice cubes quickly melt (more quickly than they would normally expect). They were able to slow the melting process by talking to their ice cube. Conversation cards were provided if the user was at a loss for words.

Bartender talks to scotch connoisseur, Mark Thoburn

Project Evolution

Here is my original design statement:

“The vast majority of the world’s population live in cities. Global issues — from climate change to deforestation — are geographically, temporally, and emotionally distant from the vast majority of us. I am interested in exploring ways to bring us into direct contact with global issues by bringing remote data sources and their associated narratives into urban settings, in physical (tangible) and public ways. This project aims to do that by injecting remote data and narratives of environmental impact into what is normally a private, enjoyable moment.”

The original project proposal for “Global IceCube” changed somewhat as I worked it up, mostly in my decision to focus on one main type of user interaction (talking), and to not incorporate a data feed into the project. More on that below.

Process Notes

This is presented in a linear form, however there was overlap between each step and considerable iteration throughout the entire process.

1. Concept Refinement and Testing:

Working from the original concept, I used simple props — a trivet, glass of scotch with ice — to bounce the idea off of people and get some initial feedback. I also spent some time with my ice cube tamagotchi, testing the melt rate of ice cubes in scotch, as well as doing several scotch and water freezing test…delicious.


User testing concept

User testing concept

Ice cube tamagotchi

Ice cube tamagotchi

Testing natural melt rate

Testing natural melt rate

Testing freezing (scotch and water)

Testing freezing (scotch and water)

2. Research:

This consisted of “expert interviews” (specific questions posed to faculty, plastics technician, data scientist) and secondary research on glaciers, possible data streams, hardware options for the heater and material options for the housing, as well as available code libraries.

At this point in the process I ended up deciding to switch from cooling the ice cube and playing around with the absence or presence of cold, to working with heat. I also decided not to control the warming of the ice cube using an external data feed, for example glacier warming data from the World Glacial Monitoring Service, various CO2 emissions sources, data specific to the user, etc. While I may experiment with that in later versions, I decided that the external context (reference to global warming) wasn’t necessary, and that the technical challenges were beyond the scope of this project. I also decided to focus on seeing how one simple user interaction  — talking to the ice cube — worked before experimenting with multiple user interactions.


3. Prototype circuit, write code and test:

In terms of code, I ended up using “standard firmata” arduino code to enable Processing to talk to Arduino, the Minim audio library , and adapted a chunk of code from Jim (Ruxton) that allowed me to use the Minim library to measure sound values when the user was talking (or not talking) into the mic. I could then set a threshold for what would be considered talking (how loud and how long) and therefore when the heater and corresponding visual feedback would be turned on or off.

The main challenge here was accounting for natural pauses in people’s speech. The first version of code I had running worked, except that it required the user to talk without pausing. I became so zoned in on the timeTalking and timeNotTalking sections of the code that I didn’t even consider digging into the minim library. An elegant solution proposed by Jim — increasing the buffer sample size for the audio to average out the pauses — helped me solve the problem.

Testing with breadboard

Testing with breadboard

Turning heater (resistor) on/off

Turning heater (resistor) on/off


4. Design/Build housing:

This was just a matter of researching materials to find one that would house the hardware and still allow for visual feedback to show through. I mapped out the physical setup of the hardware that would need to be contained within the housing, assembled a cut list and then had the good fortune to work with (assist) the plastics workshop technician, Jon Kuisma, in assembling the housing from translucent acrylic.

Housing measurements and cut list

Housing measurements and cut list

Housing (translucent acrylic)

Housing (translucent acrylic)

5. Build heater, solder circuit and assemble final installation:

After experimenting with different conductive materials for the coaster, I ended up building a quick-and-dirty heater using a 5ohm (20 watt) resistor taped to a 4’x 4′ piece of aluminium (baking pan). It took some time experimenting with different rigging systems (wire, brackets) and glues before I decided on aluminum tape, which I didn’t know existed until then, but was glad to find as it was the simplest solution, dealt well with heat, and worked.


Heating unit

Heating unit

This was my first soldering experience other than soldering two wires together. While I didn’t exactly have the hands of a surgeon, the circuit was relatively simple and I got it done with the benefit of good advice from the lab technician, George Dougherty, and helpful equipment (“third hands” are fantastic).

Here are some images of the final circuit worked up from the breadboard prototype:


Final circuit

Final circuit

Final circuit, back

Final circuit, back

6. Demo preparation and final testing: I prepared “conversation cards” (inspired by Monty Python), in case the user was at a loss for words. I also prepared a short script as an improv aid in my role as bartender. While this helped somewhat in the “performance” / demo, it’s hard work for an introvert. A true improv actor playing the role of bartender would have been better. I also spent time setting the room up and adjusting to the specific dimensions and light and sound characteristics of the room. I was able to account for differences in ambient noise and street sounds by adjusting thresholds in the code.


Code / Final Hardware & Materials List

The final hardware and materials setup consisted of:

1 x Arduino Uno

1 x 5ohm (20 watt) resistor with aluminum heat sink
1 x TIP120 transistor with aluminum heat sink
1 x 1K Ohm resistor

1 x Red LED
1 x Blue LED
2 x 330 Ohm resistor

22 gauge solid core jumper wire
3 x alligator clips
1 x protoboard

1 x audio-technica pro 24 stereo microphone
9 x “conversation cards”
1 x opaque plastic box (housing) with cutout for coaster

1 x Glenlivet 12 year-old scotch
1 x Balvenie doublewood 12 year-old scotch
1 x package of salted almonds
1 x bucket of ice

Download the code [PDF]

Overall Challenges & Successes

The group critique provided some insight into the challenges and success of this first (demo) version, although it would be useful to see what happens when testing is conducted with individual users in the intended setting — a gallery location where the installation space is small and a mostly contained/private within a public space (i.e., surrounding traffic from other gallery visitors).

The critique demonstrated that the challenges that were identified in the conceptual stage were indeed accurate. The main user experience challenges present during the critique revolved around issues of appropriate feedback and communication of interaction possibilities, specifically:

1) Appropriate feedback, specifically the changes in temperature / when the user had turned the heat on or off. I chose to use the computer screen and LEDs (changing both from red to blue) as a visual cue for when the heater was turned on or off. While the colour change was obvious during testing, the final installation location had much more ambient light; I didn’t have time to adapt and so the changes were too subtle. I think this would ultimately be a fairly easy fix using different materials and visual/aural cues. More on that below.

2) Communicating the interaction possibilities to the user, specifically that their ice cube was melting at a faster rate than usual and that they could talk to it to slow the melt rate. While I designed out a few possibilities for this, I ended up choosing to handle this through a live actor (me as bartender). I’m not sure if a more practiced improv actor would have done better, but it felt a bit clunky with me in the starring role. Communicating with the user will be one of the main challenges going forward.

3) Physical interaction challenges: the setup — a mic on a stand, no stool in front of the bar, and the heater-as-coaster — meant that the user (the serious scotch enthusiast, Mark Thoburn) could break the system, for example, choosing to hold on to his glass and not put it down on the coaster. While I realized this during late-stage user testing (and even wrote a line in my improv script to account for it), the fact that it occurred highlighted the limits of the installation and had an overall negative impact on the user experience. It also meant that the user was free to roam and so there wasn’t as cozy a relationship between the user and the bar setup as there might otherwise have been.

4) Handling variability: Mark choosing to conduct the entire class in a sing along to slow the warming (instead of interacting with the ice cubes on his own), although fun, wasn’t what I expected. Mostly because the intended user scenario was a mostly private experience within a public (gallery) setting. Still, it clearly showed that it’s impossible to fully predict how users will interact and that I need to account for that somehow.

5) The glacial pace of change: I don’t mind that the experience is meant to be longer than usual, in fact that’s what I like about it, but user expectations need to be set appropriately and then dealt with effectively. This ties back into appropriate feedback for me, but it also touches on the need to find a better hook into the experience and making sure that at some point early on, the user makes an emotional connection with some aspect of the installation, or that something gets them to keep interacting with the system.

This demo (final project) version was always intended as a first step study for a potentially larger project, for both practical reasons of feasibility — time, resources, as well as my emergent, to be generous, coding abilities — as well as the need to user test the most basic of interactions before expanding the project further.

Overall, I was happy with the process — while working with a group would have been easier, choosing to work alone forced me to get my hands dirty in putting together all aspects of the project.

In terms of successes…seeing a few people (in testing before the final demo) talking to their ice cube showed a glimpse of what I think might be an interesting interaction to expand upon. I also think there might be some potential in the theatre aspect of the demo — mixing live theatre / improv artists with an interactive scotch and ice cube installation (or interactive installations in general). Overall, I had mixed feelings about the results. Early responses to the concept ranged from perplexed to very enthusiastic; ultimately what I took away from user testing the project at various stages and the final demo/critique was that it’s a potentially interesting concept that needs some work to make it a satisfying user experience.

Future Improvements / Next Steps

One of the first things I’d like to do before taking any further steps would be some more user testing of the installation in its current form.

Overall improvements would need to deal with the challenges mentioned above. A few specific ideas for possible improvements and areas to explore (some of which came out during the critique), include:

– Bringing the user closer to the glass, in part by more closely integrating the speaking action (mic), heat source, and even feedback mechanisms (perhaps, although not necessarily by having it all self-contained). This might allow for a more intimate experience with the glass/ice, and also allow for more natural user interactions, for example, alleviating the necessity for the user to place the glass on the coaster/heater in order for the system to work;

– Research and experiment with new materials and better feedback mechanisms, including thermochromic inks for feedback based on temperature;

– Experiment with different timescales of user experience and perhaps even with multiple locations (i.e., starting the user experience in the gallery/installation and building something in that continues the experience once the user leaves);

– Explore other interactions the user could have with the ice cubes, including more physical/direct interactions, and perhaps even multiple user / group interactions;

– Explore an effective data-driven component to the installation.

Comments are closed.