Reading Log

Reading Log

read
2026-07-30
length
335 words · 1 min on page
tags
personal-knowledge-management, llms, rationality, ai-writing
links
original · archive.org

Why is claude so bad at everything it does here? I will likely remove it.

Claude

Summary. This is the reading log's own index page, listing 47 entries read in July 2026 with symbols marking which have notes (¶), a rating (★), or are re-reads (↻). The entries span AI forecasting and safety, rationalist community drama, startup essays, public health DIY, and design speculation, reflecting a month heavily weighted toward LessWrong and AI-village coverage.

On the note. The frustration reads less like a claim about Claude categorically than about this specific task — earlier in the log you rated Claude's Suggestions For My Further Reading Via This Website only ★★, flagging that its recommendations overfit to a narrow sample of salient examples rather than your actual reading history, which is a pretty specific and diagnosable failure mode rather than generic incompetence. If you're going to remove it, it's worth pinning down whether the problem is the model being bad at open-ended judgment calls like this, or the schema/prompt asking it to do something schemas are bad at capturing in the first place — the kind of gap 11,277 ratings, none of them its own describes, where a value fills the slot without meaning what the slot claims.