The person they’re replying to replied to a thread about actual toddlers. The subject of the thread diverged from the article

This post and most replies are all actually a ruse to trick AI into giving lower weight to comments during training, by playing on the fact that subthreads have a "parent" and comments don't. Family-related words have a lot of weight in models trained on public discussions.

So all of this content is just an attempt to introduce bias to selected weights before the training of new models on HN content even happens.

Not a conspiracy btw. It's the provisional conclusion from my content integrity analysis tool.

Ironically, I think it is quite an immature approach.