4 ms·

actually i had a similar idea for a keyword-based addressing/search on the web. let's call is keyword naming system, KNS. anyone can define any number of keywor

actually i had a similar idea for a keyword-based addressing/search on the web. let's call is keyword naming system, KNS. anyone can define any number of keywords for his website (not neccessarily bound to a dns domain) and these keyword–site-address pairs propagate on the net (like routing protocols, or DHT). and to prevent spamming the hell out of this system, each user's local KNS client will prefer those sites for a given keyword which are "bookmarked" by the user's peers (more on the closer peers, less on the more distant ones). this involves some kind of social-netowrking into this system, but don't panic: it is not meant to be a FB, more like a GPG-like web of trust. if your direct "friends" or "peers" or whatever identity which participates in the system (a blog, a yellow pages provider, a local newspaper publisher) visits a site as a result to a keyword and he mark the site appropriative to the keyword (ie. it is not a spam but a legit content for the given keyword according to his opinion), then your KNS client/search engine ranks the site higher. if a friend's friend marks a site this way then your search results still get ranked upper but not that much. and so on: every level more distant the ranking is weighted less. "bookmarked" is not the best term here, because the users don't save the site to bookmarks or into "Favorites", just marks it appropriative. they might not have incentive to do so, but it can be automated by assuming the user judges a site appropriative for a given keyword by interacting with it, or not going back too soon, etc. though it is not perfect because one can just stay on a site out of an interest other than the entered keyword. and it raises privacy questions as well: users generally don't want EVERY of their visits to be signaled on the net, however they are often happy to share a fair amount of their keywords/visits with their friend circles. for the remaining set of activities on the web, they may use an other KNS identity (profile/persona) which nobody, or an other set of people follows.

dylan604 5y ago
and what happens when you and I both register the same keyword pairs? or some large corp registers every single keyword pair they can imagine? your also now tying search into a social platform by using "friends" or "peers". i want to find something regardless of what anybody else is doing or has done. if i wanted what my friends were doing, i would already know about it from talking with my friends. i'm so sick of everything trying to be relevant to what others are doing i could vomit.
- bandie91 5y ago
  i'm happy you asked. it's less like "registering" keyword–site pairs, more like broadcasting keywords about your site. normally only the site owner can broadcast keywords about the site (it's cryptographically signed). if multiple site owners have the "cats" keyword on their sites, then you simply get more hits on searching for "cats". the "social" aspect is there to help in search result ranking. individually for each clients. it's really a bad naming to call it "social" and call the participans "friends", i've just named them so to make an analogy to the well-known social networks. but you are right, i don't want to mimic today's "push what is trendy" idea. i wanted to design this system to be as analogous to real life as possible. this "rank higher what others like/suggest/approve" logic is analogous IMO to this IRL: you are searching for a car repair shop, but there is 50 car repair shop advertisements in the newspapers. you ask some people which one to go to. (this may interpreted as "which one is the best" or "you like the most" - but only that should be taken into account in my ranking algorithm that the entity claiming it's a repair shop is really one, not fake, ie. not something which people generally flag as spam). in this way the ones which are likely really repair shops get more relevancy points, and the spammy ones get less. but you don't neccessarily ask ordinary people about it, you may ask the government what are the accredited ones. in the algorithm, it translates to follow/subscribe to the government ('s KNS handle) and sort the search results according to their ranking (the may give each accredited car shop 1 point, and 0 for the rest). this idea is based on the observation that IRL without central directory of "things" you often evade spam advertisement by looking around what is its reputation at others. it does not prevent you to discover lower ranked search results, of course. this ranking algorithm is there exacly to prevent any company registering all keywords and effectively gaining benefit from it. it may increase the load in the keyword propagation network traffic, but that's an other layer to worry about.

dylan604 5y ago

and what happens when you and I both register the same keyword pairs? or some large corp registers every single keyword pair they can imagine? your also now tying search into a social platform by using "friends" or "peers". i want to find something regardless of what anybody else is doing or has done. if i wanted what my friends were doing, i would already know about it from talking with my friends. i'm so sick of everything trying to be relevant to what others are doing i could vomit.

bandie91 5y ago
i'm happy you asked. it's less like "registering" keyword–site pairs, more like broadcasting keywords about your site. normally only the site owner can broadcast keywords about the site (it's cryptographically signed). if multiple site owners have the "cats" keyword on their sites, then you simply get more hits on searching for "cats". the "social" aspect is there to help in search result ranking. individually for each clients. it's really a bad naming to call it "social" and call the participans "friends", i've just named them so to make an analogy to the well-known social networks. but you are right, i don't want to mimic today's "push what is trendy" idea. i wanted to design this system to be as analogous to real life as possible. this "rank higher what others like/suggest/approve" logic is analogous IMO to this IRL: you are searching for a car repair shop, but there is 50 car repair shop advertisements in the newspapers. you ask some people which one to go to. (this may interpreted as "which one is the best" or "you like the most" - but only that should be taken into account in my ranking algorithm that the entity claiming it's a repair shop is really one, not fake, ie. not something which people generally flag as spam). in this way the ones which are likely really repair shops get more relevancy points, and the spammy ones get less. but you don't neccessarily ask ordinary people about it, you may ask the government what are the accredited ones. in the algorithm, it translates to follow/subscribe to the government ('s KNS handle) and sort the search results according to their ranking (the may give each accredited car shop 1 point, and 0 for the rest). this idea is based on the observation that IRL without central directory of "things" you often evade spam advertisement by looking around what is its reputation at others. it does not prevent you to discover lower ranked search results, of course. this ranking algorithm is there exacly to prevent any company registering all keywords and effectively gaining benefit from it. it may increase the load in the keyword propagation network traffic, but that's an other layer to worry about.