Something has been troubling me about Cliff Lynch's recent Stewardship in the Age of Algorithms and I think the source can be found early on:

By "document" here, I mean to capture a comprehensive record, or at least a good approximation, of the present reality that can be consulted today and brought forward into the future (Lynch, 2017).

Even couched in the uncertainty of approximation I think this idea of the archive as documenting reality is worrisome. Archives serve purposes. If we aren't explicit about these purposes, and instead simply talk about how effectively they document reality, then we're not really doing our job. Sometimes the purposes that documentation are put to change over time (Ketelaar, 2005). But even then it is important to know the original context, in order to put those records to new uses. In this paper Lynch moves to carve out a space that is distinct from the interests of accountability and transparency, but doesn't really describe what this space is.

Lynch also claims to be offering up pragmatic approaches several times in this article. The word is used 11 times by my count. But the key measure of the pragmatic stance (at least in the philosophical sense) is the degree to which something is useful.

Consider what effects, which might conceivably have practical bearings, we conceive the object of our conception to have. Then, our conception of these effects is the whole of our conception of the object. (Peirce, 1878)

Documenting reality is not a pragmatic enterprise. If you really want to bring pragmatism into archival work it's important to ask:

  • What purposes does the documentation serve right now?
  • What effects will the documentation have, right now?

Once we are clear on the answers to these questions then we can talk about what purposes the documentation might serve in the future. If we are always reaching for an archive that attempts to document reality then we will remain frustrated by failure, and worse, we won't really know what we are doing, and who we are doing it for.

To be fair, Lynch does go on to offer some pragmatic solutions, which are mostly pulled from Sandvig, Hamilton, Karahalios, & Langbort (2014). But Sandvig et al are specifically interested in issues of accountability and transparency. Lynch uses these methods as a launching off point, but we're not really sure where we're launching off to.

I'm as guilty as anyone for gesturing at imaginary future users who will want to know that something happened. But let's face it, it's not a very satisfying explanation of why we build archives is it? Who is the documentation being collected for? How is the documentation scoped and why? It feels like there is an undercurrent of positivism in Lynch's approach to the problem of archiving in this age of algorithms, which is at odds with the last 30 years, or more (Briet, 2006), of archival thinking around documentation and appraisal.

This blog post started when I ran across Jeurgens (2017) that put Lynch's article into relief. Jeurgens stresses the need to document algorithms but from a more theoretical standpoint. I find the connections Jeurgens makes to governmentality, sociality and the archival concept of appraisal are generative for thinking about the possible ways of scoping the problem of the archive in these algorithmic times. He rightly points out that archival concepts themselves are shifting, as they always have in response to the information technologies we use.

Mayer-Schönberger (2011) calls this process cognitive adjustment as he thinks about our understanding of permanence and loss of information in computational systems. Tracing these cognitive shifts, how they both produce, and are the product of, archival technologies is the work that needs doing. Perhaps, this is where Lynch's main contribution lies, as he talks about how the nature of computation and the presence of worldwide networks of information flows call into question the archival enterprise. Let's not assume that these archival ideas will carry forward unchanged. Carefully examining and documenting the purposes that information technologies are put to can help us understand what they mean and how they are changing.

Update: Cassie Findlay notes in Twitter that Hurley (2005)'s idea of parallel provenance has some relevance here, as we consider how archival concepts are redefined, especially in light of claims to documenting reality.

References

Briet, S. (2006). What is documentation? : English translation of the classic french text. (R. E. Day, L. Martinet, & H. G. B. Anghelescu, Eds.). Scarecrow Press.

Hurley, C. (2005). Parallel provenance (if these are your records, where are your stories?). Archives & Manuscripts, 33(1&2).

Jeurgens, K. (2017). Archives in liquid times. In F. Smit, A. Glaudemans, & R. Jonker (Eds.) (pp. 196–210). GTV Drukwerk. Retrieved from https://pure.uva.nl/ws/files/21082861/Jeurgens_Threats_of_the_Dataflood.pdf

Ketelaar, E. (2005). Archives: Recordkeeping in society. In. Charles Sturt University.

Lynch, C. (2017). Stewardship in the age of algorithms. First Monday, 22(12). Retrieved from http://firstmonday.org/ojs/index.php/fm/article/view/8097

Mayer-Schönberger, V. (2011). Delete: The virtue of forgetting in the digital age. Princeton University Press.

Peirce, C. S. (1878). How to make our ideas clear. Popular Science Monthly, 12, 286–302. Retrieved from https://en.wikisource.org/wiki/Popular_Science_Monthly/Volume_12/January_1878/Illustrations_of_the_Logic_of_Science_II

Sandvig, C., Hamilton, K., Karahalios, K., & Langbort, C. (2014, May). Auditing algorithms: Research methods for detecting discrimination on internet platforms. Retrieved from http://www-personal.umich.edu/%7Ecsandvig/research/Auditing%20Algorithms%20--%20Sandvig%20--%20ICA%202014%20Data%20and%20Discrimination%20Preconference.pdf