Data Is Not A Number
This is a re-compiled and transcribed version of a talk I’ve given at Data Kind UK&Data+Visual event at Skills Matter in March 2017. Video of the talk is available at Skills Matter website.
Note: You can click on the images to see higher res versions.
Let’s start with data. If you look at the definition of the word data we can say that data is a set of values of some kind of measurements.
With a definition like that the first thing that comes to your mind is probably a spreadsheet or a database.
Data is just the most bottom layer, the atoms from which you can construct information. By combining pieces of information together we can build knowledge.
So if we are lucky we can experience the data in a more visual way, through maps, graphs and diagrams.
But I think it's not enough so I’m interested in data that can be not only counted but also experienced.
At Variable we use a blending of design, software and aesthetic emerging from the data itself, processes and human behaviour to explore new ways of experiencing data.
I believe that data has a shape - an internal characteristic that can be extracted and exposed thought data visualization techniques and combining them with generative design systems.
I'll be talking about the shape of data in the context of algorithmic skin, a concept describing the layer of algorithms that nowadays live between our bodies and the interactions with the real world. I'm interested in the quantified-self movement myself and I find it fascinating to be able to constantly learn more about ourselves through personal data.
Couple of years ago I had a pleasure to work with Nike on a project involving Nike FuelBand. FuelBand is now discontinued but it was a device that instead of just tracking your steps it gives you fuel points for any kind of movement: walking, running, dancing, hand waving, you name it.
In this project we looked at the data of 7 female runners preparing for a race. We wanted to create "a digital award" to celebrate their efforts.
If you think about 24h day as a cycle you can represent it as a circle. We then put the fuel points you have collected on that circle next to the time of the activity (video).
This looks quite cool but I was more interested the total amount of points that you collected that day. What I did is as you collect more and more points I extrude the line higher and higher. As it grows it starts to resemble a branch of a growing tree.
When we then combine 7 of them, one for each day of the week for me they started to look like a muscle.
Next step was to look at the data of all the 7 runners together. Here we see a fuel graph for the each day of the week. Red colors represents no activity and green spikes are the times when the person was physically active. Additionally some of the days are highlighted in red as for them the minute to minute data was missing (lost FuelBand, forgot to sync etc.)
At this point I had enough data and direction where I want to go to start experimenting with the looks and different styles of rendering.
... and the long shape of the muscle fibers themselves (hence the name of the project).
Photo Source: Unknown
Working with data is a similar to the process of generative design. It's constant balancing between giving up control and falling back to curation. I had hard time choosing the right colors so I wrote an algorithm that would extract colors from Nike apparel and use those.
Here are the final visualizations. What’s interesting here is that these visualisation are true to the data. You can easily see which day was the most successful (the fiber grows highest), where was the biggest workout (thickest fiber). Does the person has a regular schedule (similar looking days) or is she more spontaneous.
You can see full video and high quality renders at http://variable.io/fibers/.
Shortly after the project has finished my father had a car accident. Nothing serious but he needed a period of recovery. He was always interested in gadgets so I put the Fuel Band into an envelope and shipped it to Poland. I hoped it will motivate him to move more just like it did for me.
What happened though is after some point he stopped using it. Watching another bunch of charts is not that exciting unless you are a data geek like me. On the mobile interface it was also quite hard to compare the days and see if you are making progress.
Taking all that into consideration I came up with the idea of a data garden. Using a similar algorithm to Fibers I displayed weekly data next to each other. This way we could see not only the daily patterns but also trends over longer periods of time. He loved the idea. Representing the data as a garden had another side effect where he started to exercise more again by caring about digital plants as a proxy of himself. I called this concept "Quantified Other" as I was learning more about him through his data.
You can see the video of the garden here http://marcinignac.com/projects/quantified-other/.
Representing data in a 3D space immediately brings a wish to be able to "be there". To visit that digital space and be surrounded by the data. I want to interact with it and be able to see both the detail and the big picture.
We had a chance to explore the idea of spatiality of data when working on the Twitter Museum Week project. The idea behind the Museum Week is to promote cultural institutions that use Twitter as their medium of communication.
There is no space without people. Therefore there is no museum without visitors.
We envisioned our visualization as a 3D sculpture representing museum responding to the followers interacting with it on Twitter. The inspiration came from the concept of Ville Spatiale series by architect Yona Friedman - a modular vertical city where it's inhabitants have influence on the shape of the space they live in.
The solid, architectural core representing the museum as a permanent institution. The number of years the museum is on Twitter twists the DNA-like spiral shape of the structure. The small modular elements are determined by profile id and the number of followers that the museum has is influencing the complexity of the whole structure. Additionally we show every tweet by the museum itself as a spherical shape inside it's structure. Every time somebody tweets #museumweek a burst of energy spreads across the museum structure and whenever somebody mentions the museum we add a stroke flowing around.
See the project page for animated versions.
Each museum had its own unique artwork available on the web available both for desktop and the mobile browsers. The artworks were customized with the colors taken from the institutions profile picture and used real-time streamed data.
To animations and more details please visit http://variable.io/museum-week/.
The digital <-> physical loop between data and place was then closed by placing the artwork in the museum itself when we exhibited it live at Cité de l'Architecture et du Patrimoine in Paris.
What was missing in the previous project was that we are still an external spectator. As I mentioned before I wanted to be surrounded by the data. That opportunity came when working on a project for Dropbox.
Here is a visualization of all the files I store in Dropbox with a color represented their type. It's mostly images with couple of code projects and quite a large number of backups both binary and text.
During this very short project (only one week) I've played with a couple of different ways of arranging the data. As this is a non-interactive piece I wanted a layout that would not only look interesting but also communicate something about the data itself.
Here is a frame from the final animation. Each block represents one file and it's scale is proportional to the file size. As you fly through the files they are sorted by how deep in the directory structure they are stored. You can see the final video at http://variable.io/timespace/.
Of course this is still a very simple and straight forward representation of the data that sparks more questions than gives answers to. E.g. how would we interact with it in VR? What other senses could we involve to "read" the data once we are immersed in it?
Representing complex datasets always comes with a challenge of explaining the data and the interface to the user. Otherwise we need do depend on lots of assumptions about his/her domain knowledge and the context. What if we could design a system that feels natural and leans on our previous experiences.
Synesthesia is a phenomenon in which experience perceived through one of the senses creates and a reaction in another sense (e.g. you can feel colors or taste sounds).
Enter Halo - a 3D data driven system that can represent patient health status, gaming performance and stock fluctuations designed for people who
monitor and interpret massive quantities of complex data in real time. Halo is completely parametric: its shape and behavior are controlled by the data and can dynamically adapt to the changing values that come from the data stream.
Here is one of Halo applications - an iOS app showing your heart rate and exercise levels. The more active you are during the day the more the Halo grows. Your current activity level is represented by complexity. Additionally we use symmetry to indicate if your heart rate is stable. This way with a glance you can get a quick grasp if everything is ok or if you should dig further into the numbers and stats.
Staying on the topic of health. In one of our recent projects we worked with data artist Julie Freeman, Dr Chris Faulkes on a project studying life and social dynamics of a Naked Mole-Rat colony at Queen Mary University in London.
Naked Mole-Rats are quite unique species as they live over 20 years (e.g. mice live 2-3 years), they almost never get cancer and are eusocial (they have a queen and workers like bees).
Here is a "state graph" of a one Mole-Rat behavior that we noted during the observations. He bit something, then walked forward, then walked backward, then he bite gain, then he sniffed twice and finally he ate something.
This visualization is one of many notations we used to capture the behavior of an individual.
What's interesting here is that this graph looks similar to a dance notation describing a sequence of steps. Maybe one day we can do a reenactment of a mole-rat using those diagrams?
There much more at http://rat.systems including info about the colony, photos of all the individuals, live tracking visualization, a data driven artwork that we use for the exhibitions and an app if you would like to see how they behave in real-time.
Using data visualization and statistics is often a good first step to get an overview of an issue. Here we see two visualizations representing wellbeing data. Left for England by Office for National Statistics and Right in London by London Datastore.
While good at communicating overall picture, what's missing is the ability to understand the situation of the individuals that contributed to that data.
I feel like this is quite common situation where you might take part in a survey or have your data automatically captured via various tracking mechanisms. That data can be then used to influence top level decisions that might later influence your life. But not often there is direct link back or a benefit for the individual that the data came from.
There is a risk that you become just a number, a statistic, an average or yet another person just like you. While in reality all that data comes from real people, with their own stories, fears, desires and challenges in their life.
Can we close the loop and design systems that benefit both data providers and data aggregators?
I had a pleasure to work as a part of a Colour-In City team trying to answer that question. Our interdisciplinary team of user researchers, designers and data scientists in cooperation with Lambeth Council investigated wellbeing of families living in overcrowded houses in Brixton, London. Through a number of workshops we tried to understand how can we make the collected data meaningful, improve their lives and create a long lasting relationship between the individuals and the public service workers. You can read more about the process on our blog https://medium.com/@Colourincity.
The result of that research is Squeezy - a friendly chatbot that tries to improve our Understanding of factors influencing wellbeing through a conversation. Every day Squeezy asks you a couple of questions about your life, neighborhood, friends and family etc. You can then reply with text, emojis or select one of predefined answers to talk about your challenges and how do you feel about that particular area of your life. At the end of the week you will get more general questions about your wellbeing and a short summary of your conversations.
There was couple of unexpected findings in this project. First the engagement from the participants was very high as they found the conversational interface very natural. The asynchronous aspect of the interaction was well suited for their busy life and many people responded to the questions send in the morning late at night when their kids were asleep.
Having a short moment of time during the day to stop and reflect on what happened was very satisfying to our participants. We decided to build on that and designed a data diary - a visual summary of your conversations. It also included a community section so you could share tips with other parents to build sense of community.
Overall during the 48 of the beta testing of our chat bot we engaged with 24 users and total of 2675 messages has been sent. Anonymous aggregated data was then shared with the service workers monitoring their community in order to influence their decisions on how to help people in need and what are the factors influencing their wellbeing.
We believe that such system proves that using data for mutual benefit is possible and it's something we would like to investigate more in the future.
As you can see we covered quite a wide spectrum of data driven systems. So how do you call those things? Is it Data Art? Liam Young and Memo Akten called it Data Dramatization, Tom Corby named it Un-visualization as an opposition to scientific and precise types of visualizations.
For me it's still an open question. As all the examples above are an expression of trying to better understand the world around us and project one of possible points of view.
So instead of data visualization vs data art I would rather talk about re-representing systems in order to foster understanding. Those representations can come at varying levels of abstraction depending on the context and what message we are trying to communicate.
If you would like to talk more how we can transform your data or just have a chat don't hesitate to drop a line to firstname.lastname@example.org.