Stract: Open-souce, non-profit search engine

Original Article ↗ Hacker News Discussion ↗

An open-source, non-profit search engine called Stract is drawing interest for its fast, uncluttered interface, rich customization options (“optics”), and features like site exploration and discussion‑focused results. Commenters see it as a rare full-stack FLOSS alternative to Google/Bing-based services, valuing its transparency, AGPL code, and potential for community influence. At the same time, many report poor relevance on common queries, missing or badly ranked official sites, and brittle matching, arguing that search quality must improve substantially to compete with established engines.

Overall impressions

  • Many commenters are impressed by the speed, clean UI, lack of clutter, and open‑source AGPL licensing.
  • Several users explicitly appreciate that it’s a full search stack (crawler/indexer) and nonprofit, not just a meta‑search frontend.
  • Others say they’ll add it to their rotation, but not as a default yet because of result quality.

Search quality and ranking

  • Mixed feedback: some find results “surprisingly ok” or even better than big engines for niche or technical queries; obscure sites and acronyms sometimes rank well.
  • Many report serious relevance failures on common tasks: basic factual queries (“how many bits in a byte”), product searches, media titles (“Dota,” “League of Legends”), calorie queries, and gaming/anime help.
  • Issues noted:
    • Stop words (“a”, “in”, “are”) appear to be weighted too heavily.
    • Matching feels too literal/case‑sensitive and brittle; synonym/fuzzy matching is weak.
    • Official/home pages often buried beneath random blogs, old pages, or low‑traffic communities.
    • Duplicated results and SEO spam surface too easily.

Features and customization

  • “Optics” (search subsets like HN‑only, blogroll, discussions, etc.) are widely praised as distinctive and useful.
  • “Explore” (similar sites from a URL) gets positive reactions.
  • Ability to block/boost sites, negative prompts, and per‑site ranking controls are appreciated; some ask to combine multiple optics.
  • Several value the clear, contextual ad model plus optional subscription.

Crawling and coverage

  • Currently relies heavily on Common Crawl; some domains, Reddit, and local‑country content are missing or poorly ranked.
  • Questions raised about how to get unlisted domains indexed and how the crawler seeds the web.

Comparisons and ethics

  • Frequently compared to Google, DuckDuckGo, Kagi, Bing, YaCy and others.
  • Some see it as “another DDG,” others emphasize that being open source and independent from Bing is a major step forward.
  • Debate around search engines filtering “misinformation”: some view such filtering as ethical and necessary; others see it as biased censorship.

UI/UX and accessibility

  • Requests for full‑width search bar, better autosuggest, easier browser integration (especially Firefox/mobile), and improved default fonts for readability.
  • Concern about default “usage statistics” tracking; some leave it on to help improve rankings but worry about popular‑site bias.

Future directions and ideas

  • Suggestions include: user voting for ranking (with spam concerns), integrating Wayback Machine content, stronger site‑weighting over time, and better multilingual support (e.g., Catalan).