Yes, my opinion is that the average highly upvoted deep dive technical article on HN was substantially better in 2023 than in 2026.
I also believe that the bulk of that degrade is due to frontier LLM’s currently being poorly trained to write such articles well, given whatever scale & sophistication of prompts the typical author is using.
I also believe that the bulk of that degrade is due to frontier LLM’s currently being poorly trained to write such articles well, given whatever scale & sophistication of prompts the typical author is using.
Not to say they won’t be better someday (soon?)