Data Humanism: my exploration of data via hand-crafted visualisations
I am about to complete a Master of Data Science and Innovation at UTS, and our final subject called on us to work on a project that aligned with our professional aspirations, and enable us to experiment and reflect on the process we undertake in Data Science. I work as a data visualisation designer, and I wanted to look beyond the current technology and tools I work with day-to-day, and experiment with hand-crafting visualisations. This is the first of a series of posts outlining my process and the outputs.
On the first day, the data is created. It could be a step down the street, the action of someone passing by a sensor, or the thump of a heartbeat. Then it is captured and stored. Then it is accessed, and merged, and manipulated. Then it is modelled, and may be pushed through machine for it to learn lessons. Then we look at its shape via a visualisation, or we may shape it into the sort of chart we like or have heard is the right way to show this sort of data to others because it’s an efficient way for them to absorb the information. And then we draw insights from this, we make recommendations, we move on to the next data set.
Often in all the backbreaking* work to get the data into the shape or the final view we need, we miss the bigger picture of what the data actually represents. What happened back on that first day, when something triggered the creation of a data point? Who or what action was recorded? Why was it captured and stored? What was included and what was omitted?
In “Data Humanism, the revolution will be visualised”, Giorgia Lupi outlines a manifesto that calls on us all to question the omnipresent nature of Data, and grapple with it at a human level. Don’t take it as infallible, but recognise it’s flaws. Don’t try to just simplify, but call out it’s complexities and contradictions. Jer Thorpe in “You say Data, I say System” calls out that a data set is an artefact** of choices made; of what to include and exclude, of who gets to be represented, of what methods are used to store, extract, combine and analyse. Data is not immaculately conceived. Rather it is created and produced within a framework that has been created by people, with biases and beliefs.
In 2015, I took the “Interactivation Studio” elective with Bert Bongers in the Design and Architecture faculty, and for my main project I came up with the concept of “Name that Tune”; a way for people to have their favourite song embedded and knitted up as a jumper that they could wear. For iLab2 I wished to keep on with the theme of data knitting, mainly as a way to keep on working with data by hand and thinking deeply about the industry I work in. Using craft also loops back to the influence the Jacquard weaving machine had on both Ada Lovelace and Charles Babbage, and the development of proto-computers such as Babbage’s Difference and Analytical engines. Knitting patterns themselves work in repetitive loops and if statements. So, I started drafting my ideas about how I could use craft to explore Data Humanism. Exploration and mind-mapping lead me back to an exhibition I saw at UTS library in 2015 called “Murder not Tragedy: Rana Plaza (A Photographic Exhibition)”. The images shown still cause me to pause and reflect on the insane system of fast fashion. At around the same time I attended a talk hosted by Oxfam about ethical fashion, and these two events made me question the role of clothing in my life. Even with wearing something so close to my skin, I often had no idea of the story of production. Since then I have become more and more reflective of my buying habits and interested in the ideas of ethical and slow fashion, largely from seeing the exhibition. This year in particular, I have made a conscious effort to buy less and make more of my clothing, and if I buy, try to be aware of origins and maker stories. I decided my project would be reflecting on fast fashion, framed against the Rana Plaza building collapse, using slow, hand-crafted techniques, such as knitting, weaving and sewing.
The complexity of the fashion system is overwhelming. Having even a simplified view of the production chain, the system is opaque and complex, due to the nature of outsourcing, the multiple elements that make up a garment; fabric production, fabric dying, thread, zippers, embellishments etc, it’s not only garment manufacturers that may be working in unsafe conditions, and not be paid a living wage. Oxfam though were at pains to point out that we should not also abandon buying clothes that have been manufactured overseas. People need work and a wage. Many of them are women. Rather, we should press our retailers and fashion brands to reveal their sources and ensure they have social, ethical and environmental policy underpinning their garment manufacturing. This issue is not black and white. It’s not just “don’t buy clothes produced overseas”.
My projects are a reaction to looking deeper into the Rana Plaza Tragedy, and exploring ways that the event could be visualised. As I made them, I was able to explore and deeply reflect on the 1129 data points, each representing a worker that was killed in the building collapse.
I had planned to create four works, but the nature of hand crafting is, well, slow. I worked for many hours and completed two, and hope to have the full four completed by December in time for “Humans, Data, AI & Ethics”. The projects are;
1. Mourning Shawl (Completed)
2. What does 1129 look like? (Completed)
3. Context (started / not completed)
4. Uncomfortable (not completed)
Each will be described and documented in the final full report.
I was interested exploring these themes further in my iLab2 project, in order to keep progressing in my own visualisation practice, and shake myself out of the “technology-tool trap”. I am a data visualisation practitioner working in a large financial institution. Data is everywhere. It is “big”. Our aim is to simplify and communicate data via trends, market movements, flows, etc. Our environment is driven by data accessibility, structure, and if it is an important indicator. Perhaps it is the last point that is the most interesting and challenging of the role. We have so much data, and only limited amount of time in the day. Our audiences are busy. These tend to be senior leaders of the business who need information delivered in an intuitive and efficient manner. We can’t make them “work hard” cognitively to get the information they need by overly complex representations of data. If they do, then we have failed in our delivery objectives. Data is used as an input to strategy, decision making and thinking. It’s important to see it as an input; it does not “drive” the business. Thankfully people do.
In this environment I am not calling on my team to ignore the important user-needs of our stakeholders, and create navel-gazing illustrative art works. Rather, I get them to ask, observe and prototype ways data can be visualised to help our stakeholders do their jobs. I have ideas though on how some of these ideas that I wish to explore in my project can be taken into my professional career and help people question the following;
- Busting the myth of the infallible data set:
- highlight, not hide issues, “missingness” or ambiguity that may be present in a data set.
- Make clear the definitions of the data shown via clear annotation and footnoting
2. Getting closer to process and systems:
- Encouraging (actually insisting) that my team has a good working knowledge about the data we are representing: how it is collected*, stored and accessed, has it been modelled and how, what processes lie behind the required visualisation, how do customers experience the process, how do our stakeholders understand their business etc
3. The joy of sketching:
- Getting data visualisation folks of all persuasions, across the organisation to down technology and pick up a pencil. The process behind sketching and making a mark, helps in organising your thoughts, and using paper prototypes actually makes us more efficient as we can discard dead ends, and bad ideas faster than if we go straight into pushing out pixels.
Next up…
*I understood backbreaking or “hard” to be the wrong way of describing what we do in data work. More to come on this in next few posts
**Lupi has also called data a medium
***There is such a huge difference between data that has been entered in manually, and by who and why, versus an automatic “ping” in the system, or if it’s a transaction, or a web-page click — and often we look at all of these data sets at once. And different storage systems mean different results too…it’s complex but it’s imperative we understand it
