I had the opportunity to put together a poster for AERI this year. The poster presents a paper that I recently gave at CSCW (Summers & Punzalan, 2017). Creating it was a surprisingly useful process of distilling the paper to its essentials while re-presenting it visually. It occurred to me that the poster session audience and the typical web audience have something in common: limited attention. So I reworked the poster content here as a blog post to try to make my research findings a bit more accessible.
Even after over 20 years of active web archiving we know surprisingly little about how archivists appraise and select web content for preservation. Since we can’t keep it all, how we decide what to keep from the web is certain to shape the historical record (Cook, 2011). In this context, we ask the following research questions:
How are archivists deciding what to collect from the web?
How do technologies for web archiving figure in their appraisal decisions?
Are there opportunities to design more useful systems for the appraisal of content for web archives?
To answer these questions I conducted a series ethnographic interviews with 29 individuals involved in the selection of web content. Participants include web archivists as well as researchers, managers, local government employees, volunteers, social activists, and entrepreneurs. The field notes from these interviews were analyzed using inductive thematic analysis.
Analysis began with reading all the field notes together, followed by line by line coding. While coding was done without reference to an explicit theoretical framework, it was guided by an interest in understanding archival appraisal as a sociotechnical and algorithmic system (Botticelli, 2000 ; Kitchin, 2016).
Coding and analysis surfaced six interconnected and interdependent themes that fell into two categories, the social and the technical, which are illustrated here in green and yellow respectively.
Appraisal in the context of web archiving is a complex interplay between the following:
Crawl Modalities: The selection strategies designed into tools and chosen by archivists in their work: domains, websites, documents, topics, and events.
Information Structures: Specific formations of web content that archivists interacted with during appraisal: hierarchies, networks, streams, and lists.
Tools: Configurations of tools that were used: archiving services, storage, spreadsheets, email, social media, content management systems.
People: Field archivists, managers, technicians, journalists, volunteers, software developers, groups (activists, professional), and institutions.
Time: How long to collect, how often to collect, how quickly web content needed to be gathered, perceptions of change in content.
Money: Grants from foundations and agencies to support collection activities, staffing, subscription fees, relationship between money and storage.
The findings highlighted sites of breakdown that are illustrated by the red lines in the thematic diagram. These breakdowns are examples of infrastructural inversion (Bowker & Star, 2000), or sites where the infrastructure of web archiving became legible.
Breakdowns between People and Tools were seen in the use of external applications such as email, spreadsheets and forms to provide missing communication features for documenting provenance and appraisal decisions.
Breakdowns between Crawl Modalities, Information Structures and Tools were also evident when archivists improvised communication tools to coordinate selection decisions when geopolitical boundaries complicated collection policies.
Breakdowns in Money, Crawl Modalities and Information Structures occurred when archivists could not determine how much it would cost to archive a website, and attempted to estimate the size of websites.
While our chosen research methodology and findings do not suggest specific implications for design (Dourish & Bell, 2011) they do highlight rich sites for for repair work as well as improvisational and participatory design (Jackson, 2014).
Thank you to Ricky Punzalan for much guidance during the planning and execution of the study. Leah Findlater and Jessica Vitak also helped in the selection of research methods. Nicholas Taylor, Jess Ogden and Samantha Abrams provided lots of useful feedback on early drafts, as well as pointers into the literature that were extremely helpful.
I also want to thank the Maryland Institute for Technology in the Humanities and the Documenting the Now project (funded by the Mellon Foundation) who provided generous support for this research. My most heartfelt thanks are reserved for the members of the web archiving community who shared their time, expertise and wisdom with me.
Noun Project images by Nirbhay, il Capitano, Creative Stall, Setyo Ari Wibowo, Agni, and Shuaib Usman Yusuf.
Botticelli, P. (2000). Records appraisal in network organizations. Archivaria, 1(49), 161–191.
Bowker, G. C., & Star, S. L. (2000). Sorting things out: Classification and its consequences. MIT Press.
Cook, T. (2011). We are what we keep; we keep what we are: Archival appraisal past, present and future. Journal of the Society of Archivists, 32(2), 173–189.
Dourish, P., & Bell, G. (2011). Divining a digital future: Mess and mythology in ubiquitous computing. MIT Press.
Jackson, S. J. (2014). Media technologies: Essays on communication, materiality and society. In P. Boczkowski & K. Foot (Eds.) (pp. 221–239). MIT Press. Retrieved from http://sjackson.infosci.cornell.edu/RethinkingRepairPROOFS(reduced)Aug2013.pdf
Kitchin, R. (2016). Thinking critically about and researching algorithms. Information, Communication & Society, 20(1), 1–16.
Summers, E., & Punzalan, R. (2017). Bots, seeds and people: Web archives as infrastructure. In Proceedings of the 2017 acm conference on computer supported cooperative work and social computing (pp. 821–834). New York, NY, USA: ACM. https://doi.org/10.1145/2998181.2998345