Since I started my PhD, I often receive questions from collaborators, friends, and family about how I discovered this research interest (especially coming from being a SWE at Stripe). The excerpt below is a short vignette describing how I decided to study tensions between craft and computing:
Stress levels were slowly building to have a submittable draft for a conference paper deadline in a few months. The postdoc (my mentor) and I were scrambling to develop textile samples and test them. It was my first year of my PhD program at Cornell in Ithaca and our project involved creating wearable textiles that demonstrate how electricity could be harvested from human motion using triboelectric nanogenerators (it is a mouthful, see our paper here for more information).
We learned and then applied a variety of textile fabrication techniques for this project (with little to no prior experience). We had just been trained earlier that year to use tools such as the Shima Seiki Computerized Knitting Machine and the Silver Reed SK155 Knitting Machine for knitting and then the floor loom for weaving (8 shaft). Learning how to use these craft tools in parallel (with their varying levels of computing and automation) led me to compare them with each other and think deeply about the emerging (and time-critical) tensions between craft practices and computational tools.
The Shima Seiki is an incredible machine. You start by designing the textile on a Computer-Aided Design software and can “print” the textile using the machine. However, I quickly discovered the challenges of divorcing the design process from the fabrication itself. There is a tacit art to configuring the design and machine settings based on the desired materials: brittle yarns required slowing down the fabrication speed to avoid breakage, elastic yarns would create pleats when added in disjoint rows and fuzzier parts of the yarn would get caught by the needles and snap. And still, for unknown reasons, the resulting fabric would still occasionally unravel as it came out of the machine. As the operator, I felt I had little control over the machine and every time there was a failure, I had to restart the entire process. It took long periods of trial and error (and discarded failed samples) to finally achieve the functional prototypes we were looking for (and even then, with some unintentional adjustments to the original design).
Although a different fabrication technique, my experience with the floor loom felt significantly different. The more senior PhD students had helped us set up the floor loom (called warping) and we learned to weave basic patterns on the loom. As we experimented with different materials, we adjusted the design and techniques accordingly. To create an imbalanced elastic sample, we needed to stretch the elastic material while weaving with elastic and inelastic materials. To do this, we used two wires with weights on either side of the comb to stretch the elastic yarn to the same level as the inelastic yarn. Whenever we made any mistakes, we were able to undo (or frog) the wefts and continue weaving. We could also cut the yarn at any point and start the next weft with new yarn if necessary. While this workflow may have been slower than the automated knitting machine, there was less material waste due to the increased agency and control we were able to exercise during the fabrication workflow.