4 ms·
We need web browsers with good local & p2p indexing/search/bookmarking features. Before I ever touch google, I want my search/address bar to look thoroughly th
by macawfish 9y ago
We need web browsers with good local & p2p indexing/search/bookmarking features.
Before I ever touch google, I want my search/address bar to look thoroughly through well organized, locally bookmarked content indexes.
The next stop before Google is indexes I've subscribed to. My friends, family, organizations, libraries, businesses, campaigns, wikipedia, etc.
After that, duckduckgo. If I haven't found it by then, google.
This kind of browser feature could make DAT/ipfs hypertext much more useful.
- QasimK 9y agoI’ve found Chrome/Chromium to be absolutely terrible at searching local history and bookmarks. It often fails to find even the simplest of pages which I know I’ve been to recently, while Firefox is capable of digging out pages that are barely in my memory. It’s one of the things that keep me on Firefox - it’s simply more usable! (Personally, I’m convinced Google does it deliberately to gain more user attention time on their search service.) The idea of having subscribed indexes is interesting! I often just want to search Wikipedia. The idea of indexes from friends or family sounds more like “trusted sources”, or the web of trust, or even GPG key verification. You’re saying I trust these sources more because I trust these people more. I think doing this slightly in the background automatically (vs actively asking your friend for recommendations, or looking at a bloggers recommended materials page) could make the internet safer. It risks creating “bubbles”, but these should be a lot less significant than your own personal (Google) search bubble.
- macawfish 9y agoYes, indeed, there is a strong risk for bubbles. In order to deter bubbles, there must still be a healthy social fabric and network applications that promote and incentivize cross-cultural interaction and knowledge sharing. That's a social, political, design & technical challenge. Also... I'm imagining the possibility "encrypted queries" between trusted parties. You could allow people to send you a search query that is encrypted in some way such that you couldn't read it, but you could still run it on your local index and send them back a result. Storing not only your own well-structured index but numerous other parties indices as well could be way too expensive. Instead, it makes sense only to maintain a relatively small local index for when you're truly offline. But to allow others to search your index anonymously, perhaps in exchange for the ability to search theirs anonymously. You could let trusted parties search your content without actually revealing their search terms, which may contain sensitive terms, like "cancer" or "schizophrenia symptoms". Maybe this could be done in a way that also obscures who actually ran the query. This sounds like a challenge for zero-knowledge proofs and some kind of anonymous traffic routing system. In practice, it'd be like a hybrid of anonymous distributed routing and anonymous distributed search.
- 5ilv3r 9y agoI had an idea for something like this. I found a fictional device called a memex on wikipedia, and I figured that it would be very easy to implement simply searching your browser cache from the address bar. Obviously the cache will need to be big and persistent, but man it sure would be useful.