The Boston view — We are what we eat (or garbage in — garbage out)
A column by Paolo Ciuccarelli
One of the most interesting and productive relationships I’m cultivating here is the one with the Harvard MetaLab, “an idea foundry, knowledge-design lab, and production studio experimenting in the networked arts and humanities”, as they define themselves.
Through my Center for Design I started collaborating with them on a project that deals with the nature of data and the processes for producing it. The aim of the project is to make the assessment of the viability and fitness of datasets for training AI algorithms easier, thus making them fairer and more inclusive. Its title — Data Nutrition — speaks for the envisioned solution: a standard label — like the nutrition labels used in the food industry — that displays the key ‘ingredients’ and measures of a dataset. The mission is as relevant as ever: the data and the visualizations we constantly look at in these days of pandemic are an example of how little we know about how data has been produced, when, why and how much this lack of information can impact on the knowledge we derive from it and how it leaves room for uncertainties that can negatively affect our perception. Beside the source — in the best case — no other information about the data (no metadata) is usually available.
No visualization can ‘cure’ bad data, and whoever is in the position to making data visible, accessible and usable should feel the responsibility of interrogating the true nature of the data as a crucial stage in the data visualization process.
Paolo Ciuccarelli
Center for Design, Northeastern University
