Re:The Vision for Decentralized Web
by megous on Sunday July 14, 2019 @06:06PM
(#20436475)
Yeah, one problem is that on the current internet, if people are not constantly posting low value crap on social media, they are hard to discover, and they're ranked down just for inactivity. But this system would support ranking just on the number of domain recommendations, and people would be able to make tools to do their own ranking (eg. how many people I recommend recommend domain X would be my rank for domain X,...), etc.
But I think, it would need to be decentralized, to avoid single DB somewhere that could be abused and manipulated, or lost.
But if you have to use a tool to read the graph from the web yourself, and people have to upload recommendations to their own server, and have them directly associated with their domain, that may encourage saner behavior, too.
If someone wants to make a centralized public search for this graph, so that people who don't see value in building their own index/ranking could search for content too, that would be possible.
But the key aspect would need to be the distributed nature. It would not need any cryptography either, just a bunch of text files at the simplest level.
The graph would be out there on web servers, the creativity would be in creating the tools to scrape it, rank stuff, and searching through it. And that could even be done in some incremental way (without the need to scrape it all at once).
But I think, it would need to be decentralized, to avoid single DB somewhere that could be abused and manipulated, or lost.
But if you have to use a tool to read the graph from the web yourself, and people have to upload recommendations to their own server, and have them directly associated with their domain, that may encourage saner behavior, too.
If someone wants to make a centralized public search for this graph, so that people who don't see value in building their own index/ranking could search for content too, that would be possible.
But the key aspect would need to be the distributed nature. It would not need any cryptography either, just a bunch of text files at the simplest level.
The graph would be out there on web servers, the creativity would be in creating the tools to scrape it, rank stuff, and searching through it. And that could even be done in some incremental way (without the need to scrape it all at once).