What this is
Social Search looks across platforms where people actually talk, then ranks your selected 7–90-day window by relevance, recency, and engagement.
No AI summary. No editor picks. Just ranked evidence: titles, snippets, links, upvotes, points, and odds.
Unlike asking an AI for an answer, you choose the window and sources, see what was searched, and can trace every result to its original post. The research is inspectable, repeatable, and easier to trust.
Searches use free, keyless source paths only. Paid-provider credentials are removed before the research engine starts. Recent repeat searches may be cached locally in your browser for 20 minutes.
How it works
Queries run through the open-source last30days engine by Matt Van Horn. Results are scored by engagement, relevance, and recency, then merged when the same story appears on multiple platforms. Dates appear only when a source supplies a credible publication date.
Sources
- Reddit — threads and comments with real upvote counts
- Hacker News — developer consensus, points, and comments
- Polymarket — prediction market odds backed by real money
- GitHub — relevant repositories and project activity, with issue noise removed
- Web — recent reporting and primary sources
- Digg — current stories clustered from high-signal technology accounts
- arXiv — recent research papers when relevant
- Techmeme — current technology coverage