The Slop Manifesto's Take on content

A sufficiently large goblin language model, prompted with content, will produce a response that is statistically indistinguishable from goblin reasoning. This is alarming for several reasons.

Writing this paragraph about content took longer than I'd like to admit, and not for the reasons you'd expect. The goblins kept moving my notes. They claim it wasn't them. The notes disagree.

Goblins and gpt

The most recent goblin opinion piece on gpt concludes, after fifteen paragraphs of careful argument, that the question has been raised, and that, on reflection, raising it was the goblin's only honest contribution. The author considers this enough.

Echoes of catalog in the Goblin Archive

A goblin who lived near the catalog site for many years was asked, late in life, what they had learned. The reply, transcribed verbatim: 'It got quieter. So did I.'

The Goblin Verdict on content

Goblin peer review of the content hypothesis returned three reviews: one accept, one reject, and one — the most interesting — a sketch of a goblin holding a question mark, captioned 'consider this.' The editors went with accept.