Over the past couple of weeks we rolled up our sleeves and gave Ecency's search engine (Hivesearcher.com) a serious overhaul. If you've used search, tag browsing, or the "similar posts" suggestions recently and felt like results were thin, stale, or not quite relevant - this one's for you. Here's what changed and what's better now.
✅ Tag search works again - searching and filtering by tags returns fresh, relevant results.
✅ Smarter "Read next" - related-post suggestions now match what a post is actually about (title, body and tags together), not just a shared tag - so they're more relevant and surface more authors.
✅ Much deeper history - search now covers content going all the way back to 2023, continuously through to today.
✅ Better ranking, three ways to sort - Relevance (default) puts the best, most on-topic matches first (and no longer gets stuck on years-old high-reward posts), Popular surfaces fresh high-value results, and Recent gives you the newest.
✅ Cleaner, safer results - spam, heavily-downvoted, and adult content are filtered out by default, with an optional "Show NSFW content" toggle if you want it.
✅ Sharper queries - exact "quoted phrases" and AND / OR / NOT operators work as expected.
✅ More reliable indexing - new posts land in search consistently, even when the wider Hive network is doing maintenance.
✅ No more duplicates - duplicate entries have been collapsed so each post shows up once.
A couple of issues had quietly crept in:
Tag-based search had stopped returning recent content. If you searched by tag - or relied on the "similar posts" suggestions that appear under an article - you were often seeing little or nothing. Posts simply weren't being matched to their tags anymore.
The search index had gaps. Coverage of older content was thin, and during a recent round of maintenance across some Hive infrastructure, a chunk of brand-new posts weren't making it into search reliably.
Relevance wasn't pulling its weight. Rankings leaned on old, high-payout posts, there was no filtering of spam or adult content, and "similar posts" only looked for a shared tag - so suggestions often missed the mark.
Individually subtle, together they made search feel a lot less useful than it should be.
Tag indexing, restored. We tracked down why posts were losing their tags on the way into the index and fixed it at the source. Tag search and the related-posts feature are fully working again.
A rebuilt, more robust index. We re-architected how posts are identified internally so that the same post is always recognized as the same post - no matter which part of the network it's read from. As a nice side effect, this collapsed a large batch of accidental duplicate entries, so search results are cleaner now.
Years of history, backfilled. We replayed and re-indexed historical content so search now reaches continuously back to 2023. Looking for something you (or someone else) posted a couple of years ago? It's findable again.
Resilient live indexing. We hardened the pipeline that feeds new posts into search so that hiccups elsewhere on the network no longer cause new content to silently slip through the cracks. Fresh posts now show up in search within moments of being published.
Beyond fixing what was broken, we took the chance to make search genuinely better:
A smarter "Read next." The related-posts strip under each article was rebuilt to match posts by their actual content - title, body, and tags together - instead of just requiring a shared tag. Suggestions are now more relevant, surface a wider range of authors, and load instantly as you read.
Smarter ranking, with three ways to sort. Search now gives you three genuinely different lenses, so switching actually changes what you see:
Relevance (default) - the best match for your query, balanced by how valuable a post is and when it was written. Substantive, on-topic posts rise above bare keyword matches nobody engaged with. Hive's early years paid out far larger rewards than today, which used to let old posts dominate every search - so we added a gentle recency balance that keeps strong older posts findable without burying everything from the last few years.
Popular - the freshest high-value posts on your topic (value plus a strong recency boost) - best for catching what's good right now.
Recent - the newest posts first, in pure time order.
Relevance is the new default on Ecency, so you land on the improved ranking without changing anything.
Cleaner, safer by default. Spam, heavily-downvoted, and adult content are now filtered out of results by default. Prefer to see everything? There's a simple "Show NSFW content" toggle to opt back in.
More precise queries. Power users, rejoice: exact "quoted phrases" and AND / OR / NOT operators all behave the way you'd expect, so you can zero in on exactly what you're after.
Search understands more than plain keywords:
Exact phrases - "hive development" keeps the words together, in order.
Exclude terms - splinterlands -giveaway finds Splinterlands posts but skips the giveaways.
Filter by author or tag - author:good-karma tag:ecency narrows things down fast.
Combine everything - "layer 2" author:good-karma tag:development type:post.
Sort by Recent, Popular, or Relevance depending on what you're after.
Hivesearcher is also a standalone service with a public API that any Hive app can build on. We refreshed the documentation and added a new /similar endpoint for "find related posts" features. Grab an API key and read the docs at hivesearcher.com/api-docs.
These improvements were made to the open-source tooling that powers Ecency's search, and the core fixes have been contributed back upstream so the wider Hive ecosystem benefits too. 💪
We're not done. We're continuing to tune how results are ranked - so the best, most on-topic content rises to the top - and exploring smarter, meaning-aware matching down the road. Tell us what you search for and what you'd love to find more easily; your feedback shapes what we build next.
Head to the search bar on Ecency and give it a spin - then let us know how it feels. If you spot anything off - a tag that isn't matching, results that look incomplete, or suggestions that seem unrelated - drop us a note. Your reports are exactly how issues like these get found and fixed.
Happy searching! 🐝
- The Ecency Team
