While Dynamicland/Realtalk and its ethos have tended to strike me as... pretentious, to say the least, the 2024 video has a key quote that really resonates with how I've been thinking about software lately:
"Anything you can make an interpreter for is a program."
Thinking of data not as something to be processed by code, but as code in and of itself, is one of those mindsets that pops up independently in various circles, but a lot of mainstream programming styles and tools and techniques seem to be borderline antithetical to that sort of mindset. I think the recent renaissance-of-sorts of AI might help contribute to making code/data equivalence more mainstream, since that seems to be how your average AI model operates: as a bunch of neurons that encode code and data as a single blob. Unfortunately, that "blob" tends to be opaque and inscrutable; I wouldn't be surprised if the next big leap in software engineering coincides with bringing data-as-code-friendly programming environments into the mainstream, such that said inscrutable blobs could be made, well, scrutable.
"Anything you can make an interpreter for is a program."
Thinking of data not as something to be processed by code, but as code in and of itself, is one of those mindsets that pops up independently in various circles, but a lot of mainstream programming styles and tools and techniques seem to be borderline antithetical to that sort of mindset. I think the recent renaissance-of-sorts of AI might help contribute to making code/data equivalence more mainstream, since that seems to be how your average AI model operates: as a bunch of neurons that encode code and data as a single blob. Unfortunately, that "blob" tends to be opaque and inscrutable; I wouldn't be surprised if the next big leap in software engineering coincides with bringing data-as-code-friendly programming environments into the mainstream, such that said inscrutable blobs could be made, well, scrutable.