> Metadata is open-world positive evidence; missing facts do not prove SQL invalid. Numeric exactness, JSON/temporal profiles, and nested containers remain separate evidence boundaries.
right
aleksiy123•16m ago
As a sidenote from the slop language, which is awful.
Feel like latest models have really got hung up on “facts” and “evidence” in a really unnatural way.
And especially Claude seems to have a lot of trouble separating it’s internal thoughts, from prompts/instructions, to output.
So you get all of these weird comments and artifacts in the output.
“neighboring versions do not inherit certification. No tag, npm publication, Pages deployment, or release authorization is implied.”
nnnnico•16m ago
api looks nice but LMAO at the readme, you should tell Claude to tone it down with the `deliberate lean-forward heavy dialect` claudisms
adzm•40m ago
right
aleksiy123•16m ago
Feel like latest models have really got hung up on “facts” and “evidence” in a really unnatural way.
And especially Claude seems to have a lot of trouble separating it’s internal thoughts, from prompts/instructions, to output.
So you get all of these weird comments and artifacts in the output.
“neighboring versions do not inherit certification. No tag, npm publication, Pages deployment, or release authorization is implied.”