The problem with this search index is that it requires all of the .torrent files containing the metadata. In lbry this metadata (as well as more rich fields) are stored to the blockchain, so any full blockchain node has the information to provide more sophisticated searches. Chainquery is what runs a full node and provides the search engine for the UI (https://github.com/lbryio/chainquery).
>As in, how would a user search it or find popular torrents?
As indicated above, if they're using the UI it provides a search engine via chainquery. Torrents are also not a currently supported source type for a download, however there is nothing to stop it from being added.
There is very basic (to be improved soon!) search functionality in the app - it searches through lbry:// urls and content metadata from publishes. Any type of data can be published to lbry, including videos, images, pdfs, and more!
In terms of how it compares to EOS, I'm not sure, but our tech is specifically designed to content publishing, disovery and monetization. It also includes publisher identities and a payment mechanism in LBC.
It parses the blockchain and syncs it with a SQL db. Anyone can run their own search server and adjust it as they see fit.
Why bother with a blockchain then? Instead of letting content creators be tipped in bitcoins or something, is there any reason other than getting to be the issuer of the coin?
Or will some future development make search decentralized?
You could also run your own search server if you don't want to use that one.
But that is only part of discovery. Another is built into the protocol. I can navigate to `lbry://bitconnect`, to see what has been published at that url. There is no way to spoof this part since it happens at the protocol layer.
Also if your interested, https://spee.ch/bitconnect. Spee.ch is a web wrapper for the lbry:// protocol. `spee.ch/url` will serve the same file as `lbry://url` (but is limited by different file types. There are more viewers in the app like a 3d file viewer.)
Anyway, great job offering working and usable software. Way better than 99.999% of blockchain projects which offer nothing but empty promises.
I could outbid that person so that the short url is directed to my channel, but that person will always have a url at lbry://@SomebodyUnpopular#xyz123...
You don't use the search server to perform resolution of claims, the search engine gives names that you then look up yourself. The resolve operations are provided by the SPV server, it cannot falsify results sent the the client as the client validates everything it gets (if the claim is winning this includes a proof that it is winning at the current block height). The server (like other SPV servers) could in some cases omit records/return false negatives, but it cannot alter them.
The SPV server is also open source and you can configure lbrynet to use whichever server you want, you don't need to use the default operated by us.
Can you explain how the same would be implemented on your system? As in, how would a user search it or find popular torrents? What would the components be and how would they fit together?
Obviously the vast majority of users (> 99.9%) will never consider running their own node, they won't need to store data on chain, they only want to read it, and they will not pay anything in order to do so.
How would your solution be better than doing it with an existing system such as EOS? It also provides publisher identities, human friendly account names, and a payment mechanism. In addition it also supports smart contracts or dapps or whatever you want to call them, which could be used to facilitate moderation.
edit: BTW, I like your demo