‘BitParts’ is the digital archive of a drawing project and handmade dataset, explored online at https://bitparts.net
Over two and a half years, from October 2019, I made a drawing, colour tile and description of 5279 things, materials, experiments and fragments from my studio. Using pencil, biro, felt-pen, and techniques associated with breaking out of habitual thinking, I modelled in a slow, analogue way what goes on when a dataset is created.
The interactive, web-based tool, BitParts, invites viewers to explore this eclectic collection of things from a workspace that often remains unseen; encountering things individually, as categories, and on a continuum from literalness to abstraction.
Whilst digitising a physical collection broadens access beyond fragile things and fixed locations, it also requires ascribing written and numerical language to the things it mediates. BitParts foregrounds and resists this translation. Through the familiar forms of drawing and things I nudge the edges of an understanding of data, how it is created and used, and suggest ways in which it can be questioned.
The website and interactions were developed in collaboration with Jen Sykes (creative coder), supported by Creative Scotland’s Open Fund for Individuals.
Taking a position of methodical irreverence, I draw on my experience as a data analyst to challenge what I see as common misconceptions around data: that it is collected (rather than created), that it is objective, that it is absolute, that it speaks. By framing drawing as data I emphasise creative, human inputs into data and the abstract thought necessary to make, and question, meaning.
The process of making the drawings and descriptions was guided by three rules:
mistakes must be visible
no new purchases
no value judgements or quantifiable information
The rules were designed to negate ideas of perfection, completeness and neutrality. They were usually, but not always, adhered to. Sometimes this was for pragmatic reasons, at others unwitting. Occasionally it was deliberate. Rather than a breakdown of the project, I see this as reflecting the inconsistencies and improvisations which are conditions of living in the world.
Mistakes were worked with, or left to exist. Avoiding new purchases necessitated resourcefulness and improvisation. This is visible in the paper drawn on, collected cast-offs such as fluorescent stock, graph paper, acetate, documentation of previous works, and pages torn from art publications. In its physical form 742 pages of drawings are contained in 19 hand-stitched sketchbooks. Compiling sketchbooks before drawing removed the temptation to discard 'failed” drawings, although the idea of failure was banned from the project as it broke rules 1 and 3.
I made the drawings using ordinary tools (pencil, ball-point, felt-pen), and techniques associated with warm-up exercises or breaking out of habitual thinking. These include fast drawing, drawing without looking, non-dominant hand drawing, over-drawing and frottage. As warm-ups these techniques are a means to an end, preparation for a more significant work. In ‘BitParts’ they are primary methods, conveying an immediacy which is juxtaposed by the extended timeframe of the project.
The colour tiles are the most 'accurate' information (recording the colour of each thing as closely as possible), yet also the most abstract. In the descriptions I use text as a drawing tool, capturing observations and responses quickly and (importantly) leaving out some information. Some are perfunctory, whereas others offer narratives or histories. Subverting accepted measures of quality, this ‘data’ acknowledges ambiguity and incompleteness, letting go of the idea of accurate representation.
A dataset is a collection of data relating to a certain topic. It can contain numerical information, text, and images. Datasets are behind
many aspects of daily life, from public transport to online shopping websites and healthcare diagnoses. They are used to build analyses that inform decision-making and policy, and in machine learning (commonly called AI) to teach algorithms what to recognise and classify.
To be usable (which means computable), data must be annotated. Information is added to describe certain details. The work of annotation is often crowdsourced through platforms such as Amazon Mechanical Turk. It is low paid piecework which demands a fast pace of working. This human input is critical, hidden, and often overlooked.
Working with data allows patterns to be seen which would otherwise be difficult to observe. The sorting, cleaning, categorising and analysis involved are processes of selection, omission, and interpretation. Whilst necessary to generate meaning, this also simplifies and smooths, stripping out variation and nuance. ‘BitParts’ refutes the idea data is objective and can ‘speak’ independently of the people who design the rules and systems which generate and interpret it. It focusses on human and creative, rather than technical, challenges to data: acknowledging subjectivity, working with idiosyncrasy, and keeping questions of means, as well as ends, alive.
The project is informed by my experiences as a data analyst, and an appreciation of fiction and puzzling.
Creative coding by Jen Sykes
Database development by Alex McCartney
Graphic design consultancy by Christiano Mere
Glossary / Elke’s definitions
datapoint one piece of information
data many datapoints
dataset a collection of related sets of data
With special thanks to Jen Sykes, Suzy Glass, Isobel Lutz Smith, Jolanta Dolewska, Priscilla Finkenauer, Hugh McCafferty, Chloe Reith, Lilias Camp and Kitty Anderson.
Supported by Creative Scotland.
/ some of the books and resources which have influenced 'BitParts'
Bassett, C., Kember, S., & O'Riordan, K. (2020). Furious: Technological Feminism and Digital Futures. London: Pluto Press.
Bellos, D. (1995). Georges Perec: A Life in Words. London: The Harvill Press.
Berardi, F. (2017). Futurability: The Age of Impotence and the Horizon of Possibility. London: Verso.
Borges, J. L. (1970). Funes the Memorius. In J. L. Borges, Labyrinths (pp. 87-95). London: Penguin Books.
Brown, B. (2001). Thing Theory. Critical Inquiry, 28(1), 1-22.
Crawford, K., & Paglen, T. (2019, September). Excavating AI: The Politics of Training Sets for Machine Learning. Retrieved from The AI Now Institute: https://excavating.ai
David, M. (2019). AI and the Illusion of Human-Algorithm Complementarity. Social Research: An International Quarterly. Vol. 86 : No. 4 : Winter 2019, 887-908.
Davis, J. E. (2019). Toward the Elimination of Subjectivity: From Francis Bacon to AI. Social Research: An International Quarterly. Vol. 86 : No. 4 : Winter 2019, 846.
Krishna, R., Hata, K., Chen, S., Kravitz, J., Shamma, D., Fei-Fei, L., & Bernstein, M. S. (2016). Embracing Error to Enable Rapid Crowdsourcing. Retrieved from arXiv: https://arxiv.org/pdf/1602.04506.pdf
Lee, F.-F. (2019). Where Did ImageNet Come From? Retrieved from unthinking photograpy : https://unthinking.photography/articles/where-did-imagenet-come-from
Malevé, N. (2019). An Introduction to Image Datasets. Retrieved from unthinking photography : https://unthinking.photography/articles/an-introduction-to-image-datasets
O'Neil, C. (2017). Weapons of Maths Destruction. New York: Broadway Books.
Perec, G. (1996). Life A User's Manual. London: The Harvill Press.
Perec, G. (1999). Things: A Story of the Sixties (with) A Man Asleep. London: The Harvill Press.
Perec, G. (2008). 'Think/Classify'*. In G. Perec, Species of Spaces and Other Pieces (pp. 184-201). London: Penguin Classics.
Plant, S. (2021). Compelled to Count. Small Black Reptile. Volume 1. 2021. Centre for Contemporary Arts, Glasgow, online at https://cca-annex.net/entry/compelled-to-count/.
Scherz, J. E. (2019). Persons without Qualities: Algorithms, AI, and the Reshaping of Ourselves. Social Research: An International Quarterly. Vol. 86 : No. 4 : Winter 2019, xxxiii.