@Konrad Hinsen we pretend that we have to read it because it's funny, according to the laws of humour discussed in the show

I don't think the definition of esolang matters. What matters is that we make the effort to find weird little languages and delight in their particularities.

Also: Jimmy's assertion that this paper shouldn't need to exist is an interesting one to consider, and we don't get to consider it without reading this paper.

Something else I am wondering about: are there esolangs that are esoteric in the original sense? Meaning small weird languages that actually have a closed community around them? Not necessarily a community that uses it for practical purposes, it could be a mainly symbolic cultural artifact.

Some APL derivates would perhaps qualify, such as J or K. And they are even used for practical purposes. And they are not funny. So... esoteric languages that are not esolangs.

People who write COBOL for banks? The COBOL CABAL.

First off, the paper you discuss is a digital humanities paper as much as the authors might not acknowledge that. Esolangs are primarily a cultural form (what they have in common is that they push back against the norms of computing), and so discussing them as a group means getting into humanities territory. You mention how much re-iterating of folk knowledge there is in some of this writing.... I agree and I wish more writing on esolangs went back to papers that have already covered the basics, starting with A Box Darkly, which introduced esolangs to the academic community early on (https://nickm.com/articles/Mateas_Montfort__A_Box_Darkly.pdf).

In terms of the "return to complexity in early coding" (something mentioned in the discussion and the paper and the first part of my paper The Less Humble Programmer... this gets to the idea (paraphrasing Wendy Chun) that "the democratizing of code is also a de-skilling." That as we moved to more English-like code, it meant giving up on the esoteric aspects of early computing where programmers would compete to shave off a few computer cycles, something valuable in the 60s but less so by the time Dijkstra got fed up with it and wrote his paper. Some esolangs, generally the Turing Tarpits, (especially early ones like FALSE, Unlambda, brainfuck), give programmers the excuse to show off technical wizardry that they can't so much in mainstream code. This is a lot of the fun of writing in these languages and is a critical part of esolang aesthetics, especially early on. (Sorry if this is all in that paper already). But we're going through another wave of this currently with llms: how will code be changed as code becomes easier to write, and what cultural forms will resistance to llms look like?

I think when LOLCODE, ArnoldC, Shakespeare, and Piet were all new, they weren't really seen as that different from each other. Each was an experiment with how strange code could look. The second type of language, where code appears as something else (a recipe, an image) is called multicoding. I interviewed David Morgan-Mar (creator of Chef and Piet) for e.c and he did not expect that people would actually make edible recipes for Chef or go nearly as far as they did with Piet designs. But they set the way for many of us who have gone on to more deliberately design multicoding esolangs. Shakespeare is a terrible multicoding language because there is almost no room for self-expression: everyones' code reads more or less the same. Chef is an interesting challenge because only certain types of recipes can work as code (mostly baking). And Piet is the canonical multicoding languages because, while most Piet programs are recognizable as Piet, they also look very different from each other, and individual programmers have developed their own visual style within it. I wrote a piece on multicoding: https://stars.library.ucf.edu/elo2020/asynchronous/proceedingspapers/17/ While Shakespeare is very limited, it inspired many natural language esolangs that are more interesting: Ashpaper, Prāsa, in:verse for a few. They picked more natural forms of expression for poets and built on those instead of the compliments and insults.

Anyway, I have a lot more thoughts but don't want to ramble too much... I think it's incredible how much you covered and how you kept it compelling for 2 hours! With the ChatGPT bits, I think there's something interesting about that ambiguity of natural language and how esolangs deal with that vs. llms (there being any single "correct" interpretation). But also I enjoy those parts because somehow it's hilarious every time Lu says "oh no."