I think the video LegalEagle uploaded explains it quite succinctly: for the sale there was a certain split between the debtors, the debtors with the largest portion were willing to forego a portion such that the other debtors would get a larger portion if The Onion’s bid was the winning one. In effect, the other debtors would get more money out of the 1.75m than the 3.5m bid, and the debtors that ‘got less’ are the ones that offered the money in the first place.

A very similar situation to that analysed in this paper that was recently published. The quality of what is generated degrades significantly.

Although they mostly investigate replacing the data with ai generated data in each step, so I doubt the effect will be as pronounced in practice. Human writing will still be included and even curation of ai generated text by people can skew the distribution of the training data (as the process by these editors would inevitably do, as reasonable text could get through the cracks.)