Hollow Search

About Hollow Search

Hollow Search is a web search engine built and run by Project Hollow, a Soulstack product. It answers from its own index rather than reselling another engine’s results, and it is the default search in Hollow Browser, Hollow Hub and Hollow’s AI agents.

How it works

Pages are collected into a corpus, then compiled into an index that is rebuilt every week. When you search, the engine scores pages by how well their title, address and text match your words, weighted by how widely each page or site is referenced, then shows a mix of sites rather than many pages from one.

For questions, Hollow AI can write a short answer from the top results, with numbered links to the pages it used. AI answers can be wrong: check the sources for anything important.

News and Images

The News and Images tabs under the search box search two more indexes of their own. News is made of headlines from 222 feeds of 162 publishers, read every hour and kept for a week: you get the headline, a short summary and a link to the publisher, never the article. Images is made of pictures on Wikimedia Commons that Wikipedia and Wikidata use: each links to its Commons page, where the licence and the author are.

Where the results come from

  • Wikipedia. Article text from Wikipedia, used under the Creative Commons Attribution-ShareAlike 4.0 licence. Snippets link back to the original article, where its authors and history are listed.
  • Popular sites. The home pages of widely linked sites, chosen from the Common Crawl host rankings and fetched by HollowSearchBot.
  • Curated sources. Documentation, references and other sites we choose to crawl in depth, also fetched by HollowSearchBot.
  • News feeds. Publishers’ own RSS and Atom feeds, read by HollowSearchBot only where the publisher’s robots.txt allows it. Publishers can ask to be removed (see the bot page).
  • Wikimedia Commons. Pictures and their descriptions from the Commons file index. The thumbnails are loaded from Wikimedia by your browser, and each picture has its own licence, shown with it.

Want it in your browser’s address bar? Make Hollow Search your default.

Independence and privacy

  • No advertising, and no paid placement in results.
  • No tracking: results pages set no cookies of their own and load no analytics or advertising scripts.
  • Queries are not logged against you; only aggregate counts are kept.
  • SafeSearch is on by default and hides adult sites and pictures. You can change it on the results page; the choice is kept in your browser only.

Read the privacy notice and terms, or report a result.

Shortcuts

Start or end a search with a bang to send it straight to another site, for example !w Wikipedia, !g Google, !ddg DuckDuckGo, !gh GitHub, !yt YouTube, !mdn MDN, !npm npm, !py PyPI, !so Stack Overflow and !maps OpenStreetMap.