What we need is search engines so easy to self-host that we can overcome the influence money & state have on Google, et al.

Like searx but less of a pain in the ass to install.

Follow

Also it federates with people's bookmarks or something? I dunno, just spitballin.

@nergalur Ooooh! Sounds like you want Odysseus!

I started it in part to explore these ideas.

@nergalur It's a web browser I'm implementing.

And perhaps the feature I'm most excited to implement is to merge the results from multiple registered (smaller) search engines into a single display, optionally including browser history. And I'm looking to aid sharing of bookmarks in order to help build those smaller search engines, entirely at the discretion of the surfer ofcourse.

So far I've implemented top sites, personalized recommendations, and webfeed discovery.

@alcinnz @nergalur
I have *ideas* about federated search, like everything else, but I'm more interested in what you're doing 😂

@yaaps @nergalur I can't say I have a full picture of how federated search should work, but I have fragments I think would factor into it.

Combined search can give you more control over who receives your queries, and bookmark sharing can help people to begin collaborating on organizing the Web.

I know curation is vital in order to outperform Google, et al, and I reckon it might be useful for these sites to provide a feed of their best links for others to harvest.

@yaaps @nergalur I do hope that by being able to focus on specific domains federated search engines can develop more powerful interfaces.

As for the interface I'm building, I'll show a specific icon in the addressbar whenever you visit a compatible site, which you can click to register that search engine. These can then be managed through the bookmarks manager.

Bookmarks would be organized by "tags" rather than a folder hierarchy.

1/2

@yaaps @nergalur You'll then be able to search your bookmarks by typing those tags (autocompleted and rendered specially) into your addressbar, or search the web by typing free-form text.

If no tags are entered then it'll send your query to all your default search engineS, and if there are tags those'll be used to filter who receives your query. The search results will be sorted by their specified "rank".

I'll help you export and share these pages.

2/2

@alcinnz
Ideally, you'd use an existing vocabulary for addressing collections of content that satisfy specific tags and also leverage an existing social graph

Ten years ago, it would've been OSB with psyc. Now, you're provably looking at SVM and ActivityPub. I don't have the coding chops to build such a thing in my mean time between mental health failures, but I believe you could

There's also some interesting discussion in SSB about communicating intent and hop count in a fine grained way that might be relevant if you'd like me to ask permission to repost here
@nergalur

@yaaps @nergalur That'd be interesting!

Though I can't promise when/if I'd tackle such a project, I've already got my hands full with two browsers and my day job.

@alcinnz
Big feels

For me the fedi is great because I see and hear about people doing the things that are unbelievably deep on my to-do. With the rare exception of someone pitching a friend on their early design solution to an NP hard problem when the friend needs something that works today and there's stuff available, I'm too happy to care about implementation details

I have ideas and I can forget about relevance of the thought and the specific expertise of who I'm talking with when I get excited, but I'm much more interested in being encouraging and helpful than inflicting my bias on others, all appearance to the contrary 😂
@nergalur

@alcinnz
My opinions lie mostly along the lines of combining social discovery, deep search, and ML classifiers

So you may use a meta search for quick results, but you'd also have communities using support vector machines to help curate ontological tag trees. The classifier files for the SVM are sharable, so it's possible to soft fork or hard fork a classifier. From there it's a matter of how you structure requests to create a sufficiently flexible, but not overly complex, privacy structure for queries
@nergalur

@yaaps @nergalur On that angle, I know an aussie who's seeking funding for generating ontological vocabularies via Machine Learning.

And I'm planning to allow "installing" these vocabularies (in SKOS format) so 1) webpages can do a better organizing themselves into your bookmarks & 2) people's organization doesn't clash as much when they share bookmarks.

SKOS would be even more important for Rhapsode though if/when I make that a full browser.

Sign in to participate in the conversation
Mastodon for Tech Folks

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!