From February 19 to February 22, I attended the annual national Code4Lib conference in San Jose, California. This is a conference for people who code for libraries, i.e., programmers, technologists, cataloguers, metadata experts, data engineers, and more. This year’s conference had broad themes around information professional ethics, privacy, machine learning/artificial intelligence, collaborative open source efforts and accessibility.
Sarah Roberts, assistant professor of information studies in the Graduate School of Education and Information Studies at the University of California, Los Angeles, delivered the keynote for this conference. Sarah spoke about the horrors of the content moderator working class. As many folks know, when you upload media to a social media provider, the media goes through a moderation process. Thousands of workers across the globe (and here in the US) are paid low wages to review this media. These workers are under a non-disclosure agreement (NDA) to not discuss the work that they do, but Sarah was able to conduct several interviews under the condition of anonymity. These workers are subjected to up-to and perhaps more-than 8 hours per day of extremely difficult content to view. Content that is deeply disturbing and distressing. Sarah’s work is unearthing this sad and alarming state of content moderation. Beyond the troubling state that deserves our attention, this keynote had the audience thinking about what other kinds of work in libraries is done in the shadows. It is important to publicly acknowledge this work.
Many of the workers in this content moderation workforce are being “hired” through Amazon Mechanical Turk. This is a service which allows you to ask someone to do any kind of discrete menial task for whatever sum of money you’d like to pay. Two other presentations discussed this as well. Jeff Gerhard gave an interesting talk on his experiments with Jupyter notebook, but also dovetailed his talk with a comparison of Centre Pompidou, and how the spirit of that building is to make a public display of the labor that goes into making a building and to honor the workers whose work is usually hidden away behind walls. On the other hand, another talk by Jason Priem of unpaywall referenced how their small company is able to keep prices down (their customers are aggregators like Elsevier and Clarivate) by making use of $10/hr Mechanical Turk offers.
Beyond this theme of exposing hidden labor, I came away with many new ideas and lessons learned. For example, I attended a preconference on the Serverless Framework and the cloud environment. Also, of particular interest to me was the MIT Libraries’ presentation on TIMDEX, a new generation of discovery that features an API-first approach. At least two talks featured machine learning applications on archival materials using spaCy. Other tech covered included VueJS, A-Frame and Go.