Show HN: AboutIdeasNow – search /about, /ideas, /now pages of 7k+ personal sites

Original Article ↗ Hacker News Discussion ↗

A new site, AboutIdeasNow, lets people search the /about, /ideas, and /now pages of thousands of personal websites to find others with similar interests or potential collaborators. Commenters praise the concept and interface, note technical growing pains (crawling, date parsing, rate limits, spam filtering), and suggest improvements such as better filtering, chronological browsing, summaries of authors, and anti-spam review workflows using LLMs. The project also sparks broader interest in standardizing personal site paths (/now, /ideas, /uses) and tying this ecosystem into social or federated networks to make independent websites easier to discover.

Overall reception & purpose

  • Strongly positive reaction to the site; many call it “amazing” and “cool.”
  • Seen as a way to discover interesting people, not just content.
  • Several users immediately add or plan to add /about, /now, /ideas, or /uses pages to their own sites.
  • Common theme: people have many ideas but lack time, execution, or collaborators; this tool is viewed as a way to find people to talk to or collaborate with.

Ideas vs execution

  • Multiple comments stress that ideas are cheap; effort and persistence matter more.
  • A key challenge is finding an idea one believes in enough to continue after the initial hype fades.
  • Some worry that publicly posting ideas may provide enough dopamine to reduce actual follow-through.

Technical implementation & reliability

  • Site initially “hugged to death”: 500 errors, DB connection limits, and overloaded Typesense instance; issues are acknowledged and fixed/upgraded.
  • Uses GPT both to infer “last updated” dates from content and to classify domains as “personal”; this leads to bugs like 1969/1970 dates and misclassified sites.
  • Fallback tooling (e.g., metascraper) and additional checks are added; users suggest JSON-LD support and HTTP precondition headers for rescraping.

Data quality: personal sites, spam, and 404s

  • Concern that many non-personal or corporate sites are included; prompt-based GPT filtering yields false positives.
  • Suggestions: allow GPT to express uncertainty for manual review; use embeddings to detect “corporate speak.”
  • 404 pages sometimes indexed; proposals include content similarity checks and stricter status-code handling.
  • Spam and later content changes raise calls for report buttons and/or GPT-based review.

Search behavior & UX

  • Semantic search is praised as “magical” for finding people who went deep on similar topics.
  • Some queries (e.g., “llm”, “gaza”) feel random or fuzzy; capitalization appears to affect results.
  • Users request:
    • Chronological or mega lists for each path (especially /ideas and /now).
    • Easier, more visible filtering between /about, /ideas, /now.
    • LLM-based summaries of /about pages alongside results.
    • Better feedback/verification when sites are submitted.

Broader ecosystem & extensions

  • Related efforts mentioned: /uses and /wish page conventions, curated lists of personal sites, and a revived web directory (Curlie).
  • Some envision this as the basis for a social network or “discovery engine,” possibly leveraging ActivityPub or IndieWeb ideas.
  • Others suggest expanding from personal sites to the wider web, while the maintainer questions why not use general search for non-personal content.