On the contrary, there's absolutely a reasoned, principled position here. Pike isn't a hypocrite for creating a Markov chain bot trained on the contents of an ancient public domain work and the contents of a single usenet group, and still complaining about modern LLMs; there's a huge difference in legality and scale. Modern LLMs use orders of magnitude more resources and are trained on protected material.

Now, I don't think he was writing a persuasive piece about this here, I think he was just venting. But I also feel like he has a reason to vent. I get upset about this stuff too, I just don't get emails implying that I helped bring about the whole situation.