Generic Gossip & Search layers

Hey guys,

Will there be generic gossip protocol functionality provided by the SAFE API that we can tap into?

I’m wondering how a search engine for SAFE would work… are there any articles on how that might work?

I was thinking vaults could run a SearchProvider program, which would be responsible for passing along search requests to their gossip network. Any node who has a strong result for a search term could return those results to the original searcher. Perhaps the person searching could pay 1 satoshi to upvote search results, which would be paid randomly to one of the searchProviders who provided the results. Other search providers could see what search terms are being paid the most and could decide to cache those results themselves…

Then you’d have all sorts of applications that would need their own search layer. E.g. SAFEinstagram etc. Users would want to search for other accounts that exist. SAFE search providers could provide application specific results, or you could even have a completely separate search layer with application specific SearchProvider nodes running a different application.

Users (voluntarily) paying for searches by upvoting would incentivize search providers to cache results and to improve the searchability of the network. Users could also voluntarily register a result with a search term if they want, which could be propagated into the search provider network - the more the user pays, the more search providers would bubble that result to the top. If it’s an open source protocol, then anyone could provide any types of search implementations they want - better implementations would yield higher payments from upvoters, likely cartels would form, but also a core group of people would run open source implementations too. Perhaps searchers could even filter the types of search providers they want to receive results from?

Not sure how to filter results to the end-user. Perhaps they could set a limit to the number of results they want to receive from each provider, and if search providers return more than that number then the user’s search program would report them and they’d be downvoted in the gossip network or something…

I’m also thinking about a business registration service, allowing people to create their own business listing, anonymously, and have people upvote/downvote/comment on them to help them build reputation, which they can stake against some service they deliver… So many applications will need a search layer, or a network gossip layer.

Am I way off the mark here? How do others think search work?

It sound like you’re on the right track to me. I’ve been discussing indexing a bit on the main forum lately too.

1 Like