This week I attended Code4Lib 2021 from my basement. This is my first virtual, multi-day conference, and I have to say I’m not a big fan. This makes me realize how important it is [for me] to go to a different physical space for a conference and do nothing but attend that conference. You want an immersive experience? Go live somewhere else for a few days and spend most of your waking hours on that experience – no dishes, no laundry, no pets demanding attention. Three afternoons spent looking at a different browser tab? Doesn’t really do it for me. [Also: you kids get off my lawn!]

For those unfamiliar with Code4Lib, it is a loosely organized group of people doing software development for libraries. You might see a presentation about developing an application at C4L one year, a presentation about implementing it at LITA Forum (sorry, that’s “Core Forum”) the next year, and a presentation about maintaining and assessing it at ALA the following year (to grossly overgeneralize).

Highlights: a very cool opening keynote by Rudo Kemper of Digital Democracy. He has worked for many years with Maroon communities in Suriname. These communities have been historically under-represented in digital archives – their oral histories seldom get recorded, and even their place names go unknown to the outside world (maps use Dutch colonial names instead). Many digital archive apps cannot be used in these communities because they rely on working Internet connections, so the challenge was to create an app that can add stories to a map, without needing a network connection to build the archive. The speaker used a local Hack-a-Thon project to jumpstart this idea, and the result is a free, open source web app available at terrastories.io. In addition to the Suriname examples, he showed projects using this with First Nations in Canada.

Other highlights include:

  • A block of machine learning sessions, including an illuminating look at PubMed’s new relevance ranking algorithm. PubMed fields an enourmous number of searches (3.3 billion in 2017, the last year with full stats), and 80% of clicks are on the first page, so there are serious ethical consequences for tinkering with the rules for deciding which search results go to the top.
  • A session on the “Mrs. Husband’s Name” problem: how to improve discoverability for people mentioned in digital records as, e.g., “Mrs. George Smith” or “Mrs. Henry Jones.” It’s a good hint to how little this can be automated that one of the software components is called “gender-guesser”.
  • A block of sessions on accessibility. Major takeaway: AI captioning is not [yet?] good enough, so fund human captioning.
  • A block of sessions on ethical computing, which took an unexpected [for me] turn into topics like the carbon footprint of big data centers and the ecological costs of storing and transmitting data over the web. I have thoughts – meandering, nebulous thoughts – that I will try to smoosh into a separate blog post.

Disclosure: I’m posting this before the actual closing session this afternoon, but there’s dishes and laundry piling up and the pets look like they’re scheming something. Good news! With these fancy online conferences, I’ll be able to catch the closing keynote later.