What Openshelf is, and what it is not
Openshelf collects research that people publish on their own websites, such as personal pages, lab pages and company research blogs, and checks that each item is a research report or a careful explainer.
Inclusion is not peer review and not an endorsement. Responses are linked so that readers can follow the discussion, not to deliver a verdict.
Points and dates are snapshots taken during the crawl. Titles and texts belong to their authors; Openshelf keeps only metadata and a one-line summary.
To suggest a post, correct a label or ask for an item to be removed: https://github.com/chenxiachan/openshelf/issues.
Sources that barely overlap
Hacker News is the traffic side: you only see what someone shared. Academic homepage feeds and Rogue Scholar are the supply side: what authors published, whether or not anyone shared it.
The 928 Hacker News rows are rule-filtered candidates, while the other two rows count every post in the window, so the share of research reports is not comparable across rows.
Topics of research reports and explainers
Each column shows shares within that source, with counts in brackets. Hover or focus a cell for details.
Whose responses get read
Each original report sits on the trunk and responses branch off it. The +N days on a response is how long after the original it appeared. Relation types were judged by an LLM from the text around each link; three labels were corrected by hand.
How this issue was assembled
What the editors left out
These sites were left out by the editors:
- jhspatial.github.io (10 posts or responses): Automated daily digest written by an AI system, not authored research.
- tobiasstenzel.com (8 posts or responses): Course study notes on data structures, not an explainer of research.
- 114514547.github.io (5 posts or responses): Personal tool troubleshooting notes, not research.
- alphaxiv.org (4 posts or responses): Mirror pages for arXiv papers, not the authors' own websites.
- emergentmind.com (1 posts or responses): Paper aggregator page, not the authors' own website.
- doi.org (1 posts or responses): Link resolves to a preprint server, not the authors' own website.
- theainewsreport.com (3 posts or responses): News-explainer site that rewrites trending posts, not responses written by researchers.
Main subject is a heated political issue (7): Elections, named politicians, partisan policy fights, migration, wars and geopolitical rivalry. Research that only touches on policy stays.
- China’s Weaponization of Dual-Use Research Vessels for Undersea Data Dominance
- Grenzmanagement statt Fluchtursachen: Wie sich migrationsbezogene EU-Entwicklungshilfe verändert
- Flagrant: Brocasting Trump, Part III
- Lost Funds: Data Update for October 15
- Lost Funds: Data Update for October 1
- Trump Tourism Effects
- I Built an AI Bot to Fact-Check the Midterms. Here's What I've Learned.