A Distant Reading Against the Code4Lib Mailing List

Since its inception, I have been maintaining an archive of the Code4Lib mailing list, and I have seen it evolve over time. In order to back up my claim, I applied distant reading to the archive, and to some degree I have been able to substantiate my observation, but only to a degree.

That said, what follows are some analyses of the archive's extracted features. To do this work I first transformed the entire archive into a set of plain text files (mailing list postings) and associated them with author, title, and date metadata values. I then modeled the files using named-entities, parts-of-speech, bibliographics, etc. I then put the resulting data set ("study carrel") behind an MCP server for the purposes of interpreting the various counts and tabulations. In the end, I have been able to thoroughly characterize the mailing list, and by extension, the Code4Lib community, over its more than twenty-year history.

Analyses

• About the Distant Reader — An overview of the underlying technology
• How big is the collection? — Actually, very big
• Keyword analysis — What is the mailing list about?
• Keywords and semantically similar words — More about "aboutness"
• Keywords and semantically similar words, visualized
• Topic model — A clustering method used to enumerate latent themes in a corpus
• Nouns — What is discussed in the mailing list?
• Verbs — What is done in the list?
• Adjectives — How are things described?
• Pronouns — Who and what are in the text?
• Dates — How have themes ebbed and flowed over time?
• Visualizations — Rudimentary charts illustrating the things described above
• Summary — A distillation of all the above
• Imaginary panel discussion — Another kind of summary
• Bar jokes — Just for fun

What I Learned

What did I learn? I learned a few things. First of all, this study carrel includes more than 46,000 individual items for a total of about 18 million words. The size of the carrel measured in words did not make my Reader application break; I have very functional carrels measuring more than twice the size of this one. On the other hand, the number of items far exceeds the number of items in any other carrel. The closest is a carrel with 15,000 items. Some of my Reader functions broke (or took forever to return). I may have to adjust something, but not right away.

Second, I confirmed the Code4Lib mailing list is a lot about library technology. It is very collaborative. And diversity, equity, and inclusion are core values. For example, the ratio between the word "he" and "she" is almost equal.

Third, based on my observations, while essentially the same as it was at its inception, the mailing list's hard-core technical threads have waned; there seem to be fewer "get your hands dirty" computer questions and an increasing number of announcements of various ilks. I don't see these hard-core issues thoroughly discussed on the Code4Lib Slack channel either, and I'd love to do distant reading against the channel's archives in order to systematically compare and contrast.

In Short

In short, as the kinda, sorta owner of the Code4Lib mailing list, it has been interesting to watch our community change over time.