For me it's two things:

1. In a dream I seem to have some ability to force myself to "wake up" if I really want to. I haven't seriously believed I could do that when I'm awake, but who knows - maybe being awake is just a much more convincing dream.
2. Sometimes when I'm dreaming there are weird enough things happening that I do suspect that I'm dreaming.

LLM generated text can also be easily detected provided you can figure out which model it came from and the weights within it. For people training models, this won't be hard to do.

I agree with the take that getting better and better datasets for training is going to get easier over time, rather than harder. The story of AlphaZero is a good example of this too - the best chess AI quickly trounced any AI trained on human games simply by playing against itself. To me, that suggests that training on LLM output will lead to even better results, since you can generate so much more of it.

A bunch of developers isn't going to do anything though if they retain control over everything. I think after learning about this background, and their weird claims surrounding it yesterday, the path forward I would prefer is for a strong fork to emerge of the original code that instances deploy instead.

I think this take makes the most sense. It seems like the totally free and open lemmy instances will do their best to re-create the Reddit that they came from. Other communities will aim for something more tight-knit (not unlike Discord servers). Both can co-exist, but it is hard to imagine the tight-knit ones taking much advantage of the federation features.