something very similar in mainframe land where they call files “datasets.” if you call a dataset a file, get ready to get an earfull!
which is ironic considering IBM themselves often call them files and one of the most popular dataset utilities is called FileAid. but if we acknowledge that, we lose a valuable opportunity to belittle and exclude newcomers





if this was about Number Go Up, they’d probably make more money selling the despined books or something instead of destroying them and selling (or do whatever they’re doing with) the raw pulp. this sounds like an expensive process and there’s no guarantee that expanding these datasets is going to improve their models in a quantifiable way, let alone a profitable one. they must just get some kind of perverse pleasure out of making the world a worse place