I may consider using HOCON [0] if I see any traction, bust after writing a lot of YAML and even making my own YAML-driven tools, I feel its shortcomings are overstated. I got bit by corner cases maybe three or four times, and they didn't take long to debug.
Perl's actually excellent at processing unstructured data, and it had a strong foothold in bioinformatics for a time. I don't think the decision was as obvious as it looks.
That's a very rosy view of humans. The average person gets older, not wiser; they see the consequences of their actions and find someone else to blame. Real social and cultural change comes from new generations.
CGP Grey doesn't explore what happens after death is defeated. I can't imagine how it can lead to anything but a stagnated society, with prejudices and inequalities enshrined forever.
De minimis is a longstanding defense in copyright law. If you are copying very little from very many works, as is the case when you turn multiple petabytes into a few gigabytes of neural network weights, you are in the clear. The problem arises when models overfit and spit out almost perfect copies of the training data.
Had they done it fifteen years earlier, I think Perl could have been the language to learn for Data Science. It still is a fantastic language to slice, clean up, and extract information from text files, and it was already very widely used in bioinformatics.
[0] https://github.com/lightbend/config/blob/main/HOCON.md