Category: Uncategorized

Aquaponic Diorama

I don’t have a ‘green thumb’­. Keeping myself alive is challenging enough, let alone sustaining a bunch of plants. But I began to imagine that these two aspirations could become linked. What if I could support my own health with hyper-local, organically grown produce? Automation would save my plant friends suffering from my haphazard watering inclinations. Using simple, cheap, physical computing components (a soil moisture sensor, a microcontroller, and a pump), I built my first aquaponic system in my small Burlington apartment.

Amazing! but then… I had to move. Moving an active aquaponic system is not a trivial endeavor. Vermont winters are harsh, for one, so fish don‘t enjoy the relocation process. Furthermore, aquariums aren’t just filled with tap water: healthy tanks are diverse, symbiotic microcosms of bacteria, minerals, algae, and tiny helpful critters. To sustain my happy aquarium, I filled jar after jar with my precious, swampy water, and carried gallons of life, piecemeal, to their new home across town.

What if that eco-infrastructure had already been in place? Aquaponics, after all, are fundamentally the marriage of plumbing and electricity supporting life. New tenants have, in the 20th century, come to expect showers, ovens, refrigerators and toilets to come pre-installed in semipermanent urban housing. Increasingly, we see dishwashers in apartments, and laundry on-site. To imagine a sustainable food future: what if, in the 21st century, we came to expect aquaponic infrastructure in apartments? What if, instead of trucking kale across the country, we could call an eco-technician and hook up to the neighborhood aquaculture grid? Maybe we already share plant cuttings, kombucha mothers, or sourdough starters with our neighbors. We’re certainly sharing the lake and the trees and the air. What about our food system?

I’m not a city planner. I’m not an electrician, a plumber, a farmer, or a zoologist. But as an artist, I had this idea: I wanted to build a physicalization of my imaginary urban eco-topia. As a girl, I grew up playing with doll houses, and then simulation strategy games of cities and civilizations. As a design student, I learned the importance of iteration, of prototyping. Great ideas start small. So here is my aquaponic diorama: a small manifestation of an idea to shift towards a sustainable food future, starting with the tiniest of apartments.

So, over the winter of 2019-2020, I got to work prototyping. The idea was to build a self-sustaining ecosystem sculpture, with watering cycles automated by an Arduino connected to soil moisture sensors and an intermittent cycle liquid pump. I tested the conductivity ranges on the soil moisture sensors with varying substrates, water volumes, and drainage systems, and built tiny planter cups to hold the plants. Originally, I used marine/aquarium epoxy to attach the drainage tubes to the planters, but the seal didn’t hold up to the manipulation required to insert the cups into the diorama. I had to redo the seals with a more heavy-duty J-B Weld epoxy.

I played around with prototypes of the water flow and housing structures… and may have flooded my kitchen floor a few times.

I built a stand, and tiny houses to go on top. Because the entire sculpture had to be wheeled into a box truck, brought to the gallery, and moved into place, I decided to build the whole system on casters. Many thanks to the Generator maker space for the use of the laser cutter to create beautiful windows and flush walls for the tiny buildings.

The first thing to be installed at the gallery was the base and aquarium. I dosed the aquarium with beneficial bacteria, and ran the aquarium cycle for a week, using substrate and some water from my home aquarium to additionally give the starter ecosystem a bit of a boost.

Wiring up the whole system was a bit of a challenge, as it required housing all the electronics in a small, watertight box inside the ‘houses’, which consisted of three layers of boxes, and threading all the wires and tubing through small holes in the walls. At one point, a small pebble from the root system of one plant became lodged in the 1/4″ tubing, causing water to overflow the planter and soak the wooden structure. Many thanks to the BCA Gallery for their patience and paper towels through all this. I created small petri-dish specimens of some of my electronic components for the education team at the gallery, to share with curious students, and illustrated the process on the walls of the gallery by converting my sharpie doodles into vinyl-cut infographics.

The show opened at the end of February, a few days after the fish and plants joined their new home. The entire sculpture was illuminated with fairy lights and waterproof LED strips, and the pump ran on a cycle every 2-3 hours, depending on the moisture level in the plant substrate. An automatic vacation feeder was loaded every two weeks with food for the fish, and evaporated water was replaced on the same schedule.

But then….

Our show “Apocalypse Diet” was put to the test when Covid-19 closed the BCA Gallery a few weeks after the show opened. My Aquaponic Diorama was designed as a theoretical model of a self-sustaining communal apartment aquaponics system, but it became fully self-sustaining when the doors to the gallery closed even to staff. For four months, the Arduino kept running, pumping water from the fish tank to the plants based on the readout from a soil moisture sensor, and the water aeration system bubbled on. The plants continued to grow and their roots soon grew out of control, snaking down the irrigation tubes and filling the tiny pots to soak up all the nutrients they could. After I was told that in the interest of public health I would no longer be able to enter the building to re-load the fish feeder, the goldfish were all transferred home. However, one plecostomus (bottom feeder) eluded all efforts for recapture. They lived in the tank, eating algae and pooping fertilizer for the plants, until an unexpected heat wave in June. The tank had been heated to sustain it in the cold, but I hadn’t considered that the air conditioning would be turned off during a global pandemic so many months later. In memory of this brave fish soul, I’ve asked folks to consider doing some reading about small-scale aquaponics and indoor urban farming, or even try building your own aquaponic system at home. The goldfish and one pleco still live with me at my home in Colchester, where I’ve been working from home for 9 months now, since our offices at the University of Vermont closed in March due to the pandemic.

As I watched the snow fall from my new home office in March, I never imagined that I would be watching the snow fall again from my home office in November. I’ve been hiking a lot, and, before the new Governor’s orders, I’d been spending time outside with friends (even watching the national election on a projector in the backyard). My office setup has changed a few times; the fish have grown bigger; I buzzed off all my hair in August, and it has already regrown almost four inches. Time passes on, and I’m grateful for my plant-filled home, for food security, for my work (which keeps my mind occupied), for the beautiful landscapes of Vermont, and especially for my loved ones, including the several family members that I’ve lost this year, may they rest in peace.

Sirius: Exploratory Analysis Python Package

Sirius is an open source Python package and web application to support a novel method for analyzing high-dimensional data using mutual information feature networks. This project is a collaboration between the University of Vermont Complex Systems Center and the MassMutual Data Visualization team. Check out and clone the project via the GitHub page here.

There are many different approaches for graphically representing feature relationships. From left to right: Small multiples (A), which are advantageous for small numbers of features of homogenous or heterogenous data types; (B) Matrices, i.e. correlation or conditional probability matrices, and (C) dendrograms, both advantageous for comparing small or medium numbers of features of a homogenous data type using summary statistics or similarity/distance measures; and (D) Networks which support a high number of features, usually of a homogenous data type, but supporting heterogenenous data types through the Sirius mutual information implementation.

Mutual information scores allow us to find feature pairs which are highly dependent. Feature pairs with high mutual information scores are connected in the resultant network graph.


The tool is designed to process data of continuous and discrete data types; mutual information can be calculated among homogenous or heterogeneous data type pairings. Unlike correlation, mutual information allows us to find dependence among features which is non-linear.


A backbone method is applied to the feature network which selects the most statistically significant edges and displays them in the resultant web-based application.

By thresholding the network in a context-aware manner, clusters of dependent features remain in a simplified network visualization view. Compare, below, the statically thresheld association network (left) with the dynamically thresheld mutual information network generated by Sirius (right).

These networks show great potential for EDA of high-dimensional data sumarries, while still allowing direct access to pairwise record-level visualizations.

      Stay tuned for the paper.

Inside, Outside: Installation at Champlain Maker Faire

“Technology made it easy for us to stay in touch while keeping our distance;
’Til we just stayed distant and never touched. Now all we do is text too much.”
– Sage Francis, “The Best of Times”

I find myself saving images of campfires on Pinterest… I bookmark hiking trails on my smartphone… I go to a box store and pick out dead flowers under fluorescent lights and carry them home, like a house cat with its prize. “I should be outside” I think as I scroll mindlessly though Twitter. Real birds don’t tweet about hashtags. I daydream of the forest when I sit in my living room — so now, I’ve brought my living room to the forest.

Our Compost Ecosystem: Breaking Down Vermont’s Food Scraps

By Jane Adams

Every year, the folks at MassMutual Life Insurance get together for “Data Days for Good”, where team members work to help a local agency or organization with a data science project. Through a research partnership between the University of Vermont Complex Systems Center and MassMutual, I learned about the Data Days for Good initiative, and in the spirit of that event, I turned a data artist’s eye toward our state’s evolving compost ecosystem.

As a gardener, I know the benefits of composting food scraps for improving soil quality and plant nutrition. What I didn’t know: there is a real cost to not composting. I had always assumed that food that didn’t make it into a compost pile would simply decompose in a landfill, but that’s not the case. Food that would decompose in a matter of weeks in a compost pile can take years to decompose in a landfill, taking up space and releasing methane into the atmosphere. Take carrots for example: In a backyard compost pile, they will be unrecognizable in just 9 weeks:

0 weeks 4 weeks 9 weeks

In a landfill, because there are no decomposers in this anaerobic environment to digest the material, carrots that have spent
nine years sitting under trash still look like this:

Many thanks to Lauren Layn at Green Mountain Compost for these fascinating photos, and for teaching me a whole lot more about compost in our great green state.

There is only one landfill in Vermont. It’s located in Coventry, just south of the Canadian border, and last year Casella Waste Management processed around 420,000 tons of waste there from across the state. The landfill is beginning to reach capacity, and while there are discussions taking place to address where future landfill materials should go, there is a more immediate solution: decreasing the amount of recyclable and compostable materials that end up there in the first place.

A 2012 study from the Department of Environmental Conservation found that 28% of the state’s municipal solid waste was organics. Paper, plastic, and glass combined to comprised another 35% of what was being sent to the landfill. This data was part of the impetus for Act 148, the Universal Recycling Law, instated in 2012.

With the enactment of Act 148, Vermont became the first state to ban all food scraps from landfills. Setting a precedent doesn’t come without some logistical challenges, though. Like many of the systems we study here, this one is complex. Lawmakers in Vermont recognized this fact, and created a timeline to gradually phase out food scraps from landfill streams, beginning with large-scale commercial waste generators in 2014 and expanding the same requirements to households in 2020:


2014 2015 2016 2017 2020
Food scrap generators of 2 tons/week must divert material to any certified facility within 20 miles Food scrap generators of 1 ton/week must divert material to any certified facility within 20 miles Food scrap generators of 1/2 ton/week must divert material to any certified facility within 20 miles Food scrap generators of 1/3 ton/week must divert material to any certified facility within 20 miles Food scraps are banned from the landfill and haulers must offer food scrap collection
Transfer stations / bag-drop haulers must accept leaf and yard debris seasonally Leaf, yard, and clean wood debris are banned from the landfill Transfer stations / bag-drop haulers must accept food scraps


via the Act 148 timeline, available here.

For curious folks in Vermont, the Agency of Natural Resources (ANR) has a fantastic “Open Data Portal” where you can find data sets about everything from fisheries and geology to waste facilities and food scrap generators. Using this data, I was able to start making sense of just how much food waste we produce as a state. Granted, a lot of these values are imputations from self-reporting and estimates based off categories like facility type (restaurant, school, hospital) and metrics (number of seats, students, beds), but the ANR put a tremendous amount of work into cataloging and making a best guess at what the actual food waste output is for over 5,000 companies and institutions across the state. When we put these places on a map, we see food scrap generation clustering around population centers and major roadways:

(Mouse over the canvas and scroll, or use the ‘+’ and ‘-’ buttons, to zoom. Click and drag to pan. Click on points to learn more.)

A first glance at the data shows that our distribution (the number of places generating food scraps across different categories of volume) skews very far to the left: many places produce a small amount of tonnage per week, while just a small handful generate a tremendous amount of food waste every week:

(Click and drag to select an area to view; double-click to reset the chart).

Actually, if we hover over the very tail end of our histogram, we’ll see that there is just one place producing a whopping 288 tons of food scraps every week. Along with maple syrup and beer, Vermont is known for its cheese, so it’s no surprise that our outlier, the state’s largest producer of food waste, is the Cabot farm in Middlebury, Vermont. As it turns out, their food waste goes right down the street to Foster Brothers Farm, where owner Bob Foster (who is also Chair of UVM’s College of Agriculture and Life Sciences Board of Advisors) has partnered with local green energy innovator AgriLab Technologies to create a thermal power system that uses the heat of decomposing waste to speed compost production and generate energy (read more about their story and others from Seven Days here). Their compost, “Moo Doo”, is sold to gardeners all over New England.

People are often surprised by how much heat large compost piles generate. From the photo below, taken during a visit to Green Mountain Compost here in Chittenden County, industrial-scale compost piles are required to reach at least 131°F, so it’s no wonder that folks around here have started harnessing that powerful thermal energy in the cold winter months.

From the ANR’s data, we find that estimated non-residential food scrap generation across the state totals around 2,500 tons per week. That’s 5 million pounds every week, or 260 million pounds every year. Of that total, about 13% of all estimated food scrap generation is from just two places: Cabot Creamery in Middlebury (288 tons/week), and West River Creamery in Londonderry (49 tons/week). Vermont sure loves cheese! When we remove these two outliers from our distribution, we get a slightly less skewed result, with the majority of non-residential facilities estimating less than one third of a ton of food waste per week:

(Click and drag to select an area to view; double-click to reset the chart).

What does this mean for our state meeting the 2020 goal of banning all food scraps from landfills? Well, since 2012, based off of the estimates we have and the benchmarks set by the law, we might have expected to see tonnage per week and total number of non-residential generators composting increasing by these amounts each year:

This reflects what we would expect, knowing that high-tonnage food scrap generators are few but large, and lower-tonnage generators are numerous. On a map, we can see food scrap generators that are currently required to compost under Act 148:

Come 2020, however, the number of food scrap generators (which will then include residential generators as well) will increase significantly:

Here is the same map as above, with the addition of the remaining non-residential food scrap generators that will be expected to compost their food scraps by that time:

If we isolate the new additions and aggregate them by locality, we can see from the hex-binned map below where we might expect an increased demand in food scrap collection:

(Note: Totals listed for each hex bin indicate total number of facilities that will be expected to compost by 2020 being added to those who are already composting in that area).

So, who is collecting all of these food scraps? Where do they go? The Agency of Natural Resources has created an interactive Materials Management Map, available here that allows local residents and business owners to connect with food shelves and food scrap waste facilities in the area. Additionally, there is a regularly updated directory of food scrap haulers across the state, with areas that they serve.

One important thing to note about Act 148 is that it’s not just about sorting waste: it’s also about conserving and diverting materials, so that less goes to waste overall. The Vermont Food Recovery Hierarchy, based off of the EPA’s guidelines for managing food waste, emphasizes that decreasing food waste starts with reducing overconsumption to begin with. Then, food that is still fit for human consumption should go to people.

I spoke with Natasha Duarte of the Composting Association of Vermont, and learned from her about Vermont’s “charity food system”. The Vermont Foodbank  cites a statistic that “in Vermont, 1 in 4 people struggles with hunger”. According to Natasha, since the establishment of Act 148, the charity food system in Vermont has seen an estimated 40% increase in donations to food shelves.  Salvation Farms is one such member of the charity food system here in Vermont. That organization provides job skills to workers, such as youth and prison inmates, who glean (rescue un-harvested produce) from farms across the state and deliver the food ‘seconds’ to local food shelves.

Image below from VPR: “Prison Labor Bringing Gleaned Crops To Food Shelves”. 

The Vermont Farm-to-Plate Network Atlas allows users to find everything from food producers (farms) and consumers (restaurants, food shelves, businesses) to waste facilities, community gardens, and compost producers. Between that resource and the ANR’s Materials Management Map, it’s easy to see how informed Vermonters might get on track to meet the 2020 composting deadline. Check out those resources, as well as this map showing aggregate binning of non-residential food scrap generation alongside waste facilities and community gardens (enormous thanks to Libby Weiland at the Vermont Community Garden Network for sharing this data):


Meanwhile, here in Chittenden County, residents and business owners are stepping up and organizing ahead of the 2020 goal. I spoke with Cameron Scott and Jacob Wollman, who run No Waste Compost, a compost pickup service in the area. Their mission is “to make an affordable and responsible composting service for all of Vermont”. Their “smell of the month” videos are something of a hit among customers:

The No Waste folks bring their residential pickups over to Green Mountain Compost in Williston, which is a facility run by Chittenden Solid Waste District (CSWD). Haulers can drop off food scraps there for a ‘tipping fee’, which is determined by the weight of the waste dropped off.

Composting isn’t just for food scraps! Below, we see a recent deposit of leaf and lawn debris at Green Mountain Compost from a local landscaping company:

Green Mountain Compost offers composting advice and tours of their facility to the public. I learned a whole lot during my time at their “Edu Shed” and touring the facility. You can sign up to take a tour here.

Above: At the Green Mountain Compost “Edu-Shed”, where tours begin and we learn about why and how to compost.  Below: Alyssa Borowske (left) of the Department of Environmental Conservation; Lauren Layn (center) of Green Mountain Compost; and me (right), excited about the future of Vermont’s compost ecosystem!

Another approach under consideration? Methane food digesters. Alex DePillis with the Vermont Department of Agriculture explains: “Taking food waste to digesters is a pathway we’re working on in Vermont.  So far, we have one example: Pre-ground Norwich University cafeteria and catering waste plus brewery waste from Alchemist and others [was] combined [and], in a vacuum truck by Grow Compost, hauled to Vermont Technical College, who then spread the digestate (liquid agronomic nutrients from the digesters)”. Read more about “Big Bertha”, Vermont  Tech’s anaerobic digester, here.

If you haven’t begun composting at home already, now is the time to start! Haulers across the state are increasingly accepting food scraps and yard debris from residences, or you can look locally for a community garden where you can contribute to the production of that ‘black gold’ compost that makes our farmed food so delicious!

Many thanks to the awesome folks at these organizations for teaching me more about Vermont’s compost ecosystem:

especially to Lauren Layn at Green Mountain Compost, Natasha Duarte at Compost Association of Vermont, Alex DePillis at Vermont Agency of Agriculture, Emma Stuhl at the Vermont Agency of Natural Resources, Libby Weiland of the Vermont Community Garden Network, and Cam Scott and Jake Wollman of No Waste Compost.


Further resources:

Vermont Department of Environmental Conservation

Vermont Agency of Natural Resources

Vermont Agency of Agriculture Food & Markets


Vermont Farm-to-Plate Network

Vermont Community Garden Network

Vermont Foodbank

Hunger Free Vermont

But What About…?

“But What About…?” is a prototype of an exhibition that explores the process of cartographic data visualization to answer for young people (ages 18-35) the question of where to live. In essence, it is an exhibition that curates itself, as attendees are invited to ask their own questions about which metrics inform our assessment of livability.

The space, which is purposely modular and informal, provides an area for young people to study the questions that were posed by graduate students Jane Adams (MFA), Brent Burns (MS) and Serdar Gizer (MS) in the Emergent Media program, and to get an impression of the challenges they faced in trying to answer those questions. The idea of an ‘exhibition prototype’ opens up a whole realm of possibilities for future iterations which further emphasize the relationship between humans and data, quantitative and qualitative metrics, and which factors are subjective or universal criteria of livability for young people.

If you are interested in hosting a future iteration of this exhibit, please contact

A Serious Game

“A Serious Game”, inspired by an assignment of the same name, is an interactive art experiment aimed at bringing to light the potential for news feeds to be surreptitiously augmented. The piece would be housed in a number of public spaces and controlled by a kiosk and/or mobile application.

Coded in Python, any number of users could be simultaneously augmenting the feed at any given time by using a ‘mad libs’ style find-and-replace feature to change keywords in the feed for satirical, confusing, or subversive means. The feed would pull fresh RSS news headlines from the web every 30 seconds, such that new unaltered information would be blended in with the augmented news, blurring the line between reality and crowd-sourced nonsense. This piece is inspired by Facebook’s ‘News Feed’, Twitter, and issues of truthfulness in the modern mediascape’s echo chamber.

Algorithmic Chromo-synesthesia


My latest side project: color-encoding music scores with palettes assigned according to the circle of fifths, and building networks of songs from their midi files using Mido & Plotly in Python.

Check out the notebook here and the GitHub here!

Throughout history, composers (and many others) have described a phenomenon in which musical notes manifest as colors in the mind’s eye. Consider this chart by Fred Callopy showing three centuries’ of recorded synesthesia:

While I don’t personally experience synesthesia, I am a very visual thinker, and in self-teaching guitar I wondered at the potential of this system for improving learning. By studying the commonalities in the above chart, I created this system for color association:

I then painted the fretboard of my guitar according to these colors, which produced a beautiful artifact:

Then, I started mapping the colors to sheet music. Here are some of the results:

“Frère Jacques”, where red = C, yellow = E, blue = G, etc. and length of rectangle indicates length of note, such that a square is a quarter note:

Below, “Ode to Joy” introduced problems of identifying adjacent octaves; the low G in row 3 is differentiated from the higher G in other rows by a decrease in brightness, such that hue still corresponds to note but brightness corresponds to octave:

Unlike “Frère Jacques” and “Ode to Joy”, which are very bright rainbows, the color palette generated by sheet music for “Fur Elise” is decidedly unsettling, much like the tune itself:

Note the very light yellow note in the 5th line; this is the very high E, as higher octaves are identified by increased brightness in the same way that lower octaves (such as the mustard-colored low Es preceding it) are darker.

Here are some experiments I’ve done trying to perform pitch detection; first using FFT as part of Processing’s minim library, then using the Google Teachable Machines audio software (still in beta testing):

WARNING: This video starts with a very high-pitched sound


Interactive Terrarium

In order to foster creativity of a tactile nature, a daily practice was developed in which elements were arranged into small sculptures over the course of ten-minute sessions. Initial specifications required that objects be sourced from the immediate environment, and consist of at least one man-made and one natural element, with no less than three total elements per sculptural composition. This promoted aesthetic dialogue between the organic and the synthetic, echoing a dichotomy that has deep cultural resonance for the millennial generation.

Creative blocks in the daily practice led to a more holistic approach, which abandoned the requirement of small, ten-minute complete sculptures in favor of an ongoing study of physical computing and terrarium cultivation. The result was a culmination of the semester’s research into a large-scale interactive sculpture incorporating both live plants and live circuits. This piece was on display in the Champlain College Communications & Creative Media Gallery lounge as part of the graduate student exhibit “10×105: Explorations into Daily Creativity” in December 2016.

Stalking the Iron Mule

This interactive spatial narrative was created for a digital storytelling course. It is an experiment in a non-linear take on the oral tradition. Since footage is in excess of 3 hours, it is near impossible to experience the story in its entirety. Instead, users are encouraged to jump around the map at their leisure, moving forwards and backwards in time, much in the way that the subject’s own memories are incomplete and achronological. The narrative is bookended, however, with introductory and closing videos that provide some deeper context to the content in between. You can experience the piece in full screen mode here. As noted in the introduction, this is a work of FICTION. All similarities to real persons, places, or events are purely coincidental. All footage is shot in Vermont.

Good Filth

In a country where over-prescription of drugs is rampant and the “quick fix” for mental and physical ailments comes in the form of a pill, horticultural therapy is an emerging field that seeks to bring people back to the earth as an alternative coping mechanism. Horticultural therapists propose that the act of growing plants, from digging ones’ hands in the soil to sow seeds, to the satisfaction of a harvest, can have profound benefits to overall health.

Many students and young professionals do not start home gardens because they perceive the process as complex and expensive, so it is important to adopt a design strategy that conveys simplicity and cost efficiency. This can be achieved with a two-pronged approach: first, repeated symbolism of step-by-step processes, breaking complex instructions into small tasks that are manageable to a first-time gardener; and second, a DIY style of imagery that conveys the importance and cost-saving advantages of repurposed materials.