IdeaWell

Project ideas from Hacker News discussions.

Show HN: I used Claude Code to discover connections between 100 books

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. Superficial or Meaningless Connections

Many users criticized the LLM-generated links between books/phrases as weak, arbitrary, or hallucinatory.
"Aurornis: "The visual style... looks neat, but the connections don’t seem correct."
"smusamashah: "the connected words have absolutely zero connection with each other."
"timoth3y: "it's an LLM Rorschach test where we are given random connections and asked to do the mental work."
"tmountain: "The links drawn between the books are 'weaker than weak'."

2. Distinct LLM Style and "Slop"

Commenters immediately spotted AI hallmarks in descriptions and outputs.
"Aurornis: "many of the descriptions have a distinct LLM-style voice... I would have immediately recognized many of the motifs."
"pmaze: "The names & descriptions definitely have that distinct LLM flavour."
"typon: "Just standard LLM slop once if you actually have read some of these books."
"drakeballew: "egregious, obviously referential LLM dog."

3. Value as Innovative Tool for Discovery

Praise for the concept's potential in "distant/syntopical reading," UI, and augmenting human insight, despite flaws.
"pxc: "distant reading... zoom out to hundreds... using computers to search... insights that only emerge from large bodies."
"Balgair: "This is his Syntopicon for modern works... Brilliant!"
"znnajdla: "It takes a very high degree of artistic creativity... Every one of these connected threads are really good."
"sciences44: "Love the originality here - makes you curious to explore more. Solid technical execution."

🚀 Project Ideas

Syntopicon AI: The Modern Great Conversation

Summary

  • A "Syntopicon" for the modern age that helps researchers and readers find thematic links across thousands of books and papers, inspired by Mortimer Adler’s The Great Conversation.
  • It solves the "semantic gap" in research by moving beyond simple keyword search to "distant reading"—linking specific concepts (e.g., "OODA loops" in fighter pilot training) to analogous concepts in seemingly unrelated fields (e.g., "Kitchen Confidential" and kitchen efficiency).
  • Core value: Automating the discovery of hidden conceptual threads across a massive library.

Details

Key Value
Target Audience Researchers, students, and "Syntopic" readers (e.g., users like Balgair and pxc)
Core Feature Thematic "trails" connecting excerpts from different texts based on shared underlying logic
Tech Stack Python, PostgreSQL/pgvector, Claude API (Contextual Retrieval), D3.js for visual linking
Difficulty Medium
Monetization Revenue-ready: SaaS subscription for personal knowledge management collections

Notes

  • HN commenters specifically mentioned Mortimer Adler’s Syntopicon: "This is his Syntopicon for modern works, and automated. It's amazing..."
  • Addresses the criticism that "100 books is too small a datasize" by allowing users to import their own massive PDF/EPUB libraries (mentions of 1000+ files).

Code-to-Folklore: The "Story of Mel" Analyzer

Summary

  • A specialized tool for developers and historians to uncover the "art" and "personality" hidden within complex, undocumented, or arcane source code.
  • It uses LLMs to "close-read" code not just for functionality, but for artistic intent, quirky optimizations, and the developer's "voice"—similar to how the "Story of Mel" describes uncovering the genius in hexadecimal code.
  • Core value: Making "incomprehensible" GitHub projects accessible by explaining the domain context and the "why" behind weird implementation choices.

Details

Key Value
Target Audience Developers, technical archeologists, and open-source contributors
Core Feature "Aesthetic" code commentary and "Why this is brilliant" deep-dives for legacy/complex repos
Tech Stack Claude Code (agentic workflows), Tree-sitter for AST parsing, React
Difficulty Low
Monetization Hobby (Potential as a GitHub App for education/documentation)

Notes

  • Inspired by the discussion on whether code is art: "I have always found the story and 'pessimal' instructions beautiful... Kaye and Nather are both artists to me."
  • Solves the frustration mentioned by dinkleberg: "There are so many interesting projects on GitHub that are incomprehensible without a ton of domain context."

Narrative Rec Engine (NRE)

Summary

  • A movie and book recommendation engine that ignores genre/tags and instead uses "narrative resonance" and structural similarity (3-act structure, pacing, character arcs).
  • It fixes the failure of current recommendation systems (like Netflix or Goodreads) which often suggest books/movies based on shallow associations rather than the "feeling" or structural arc of the story.
  • Core value: Finding content that fits a specific "vibe" or narrative structure (e.g., "stories about secrecy where the protagonist is their own worst enemy").

Details

Key Value
Target Audience Power readers, cinephiles, and users tired of "similar item" algorithms
Core Feature Structural fingerprinting (pacing, themes, structural beats) to find narrative matches
Tech Stack Python, OpenAI/Anthropic APIs, Pinecone (vector DB), Flask
Difficulty Medium
Monetization Revenue-ready: Affiliate model + "Pro" filter features

Notes

  • Directly addresses hising’s pain point: "used Open AI API to break down movies into the 3 act structure... identifying movies based on more than 'similar movies' in IMDb."
  • Solves lkbm's "portal fantasy" problem where traditional lists/tags fail to categorize books by specific narrative elements (e.g., "portals located on water").

Read Later