In general, the article will discuss how, guided by the tastes and talents of users of the platform golos.io/steemit.com, get information that will appeal to the audience. The proposed rating is based on the idea of ​​a peculiar delegation of “tastes”.

Who wants more details about the platform Voice - look here or, in three words, look here . Steam and Voice are brothers, there are nuances, but they are not important to us.

But to read and understand this article, it will be enough to know that golos.io/steemit.com is a blogging platform (based on the GOLOS / STEEM blockchain, i.e. with payment in cryptocurrency, but this is not so important) and that they (or rather, in the GOLOS / STEEM blockchains, but this is also not so important) for "some" user actions (for example, for publishing an author's article, but a little more complicated) accrues reputation, which we take as a basis.



Those. the essence is that there are some activists of the platform, and their overall reputation can be considered a qualitative indicator of their activity and success .

Now to the essence of the very idea of ​​selecting articles (ie rating)

So, we have blogger users who publish articles using certain tags. Today, the daily user activity brings at least 1000 articles to each platform, and nobody wants to read everything, just as you don’t want to miss something interesting.



Those. to physically cover such a volume is impossible (for me personally, for sure, but perhaps there are also “informational freaks”, who knows?).



But, if you think about it, with the right approach, we don’t need it.



The line of thought is as follows: if a user publishes some material with a tag, for example, “engine”, and gains likes from local activists, his reputation grows by the tag “engine” + the general reputation of the user. Those. if many people like the article, then its reputation in this tag will increase significantly. If the article comes out "so-so" in the opinion of users (= like a small number of people), then the reputation, though growing, is insignificant. And if you don’t like it, the author’s reputation may even go to the “minus”.

Those. In general, with the proper level of reputation, it turns out that such a "national" specialist is precisely on the tag (read - the topic) "engine".



In turn, the “people's” specialist also reads someone's articles, rewarding some of them with their likes. Giving his “voice” (= Like) an article on the theme “engine” in which he acts as a kind of recognized expert, the “people's” expert signals us that she is interested, in his opinion. Suppose that we have more than a dozen or more of such "popular" specialists in the "engine" tag - it turns out that based on their taste, you can create a list of recommended content for other users.



This is the main idea and the whole "salt" of the rating.

So let's do it again briefly.

• each user has a common reputation
• for each user reputation is calculated by tags
• for articles the rating is calculated by tags
  
  
  
  Those. overall user reputation → reputation by tag → article rating by tag

Example

• Peter published an article with the tag "food", he liked Vanya with a reputation X. Hence, Peter now has X reputation for the "food" tag , and the value of X will be taken as an indicator of his ability. At the same time, in all other tags, Petya has zero reputation (that is, just like on Habré, there is no article - there is no vote).
• After some time, Lena publishes an article with the same tag, "food", and gets likes from our Petit. Petya has Y's reputation as a user, and she will go to Lena as Y's reputation on the "food" tag , and Petya has a reputation on the "food" tag X , and this reputation X will go to Lena to rank the article on the "food" tag . Those. article rating is formed by the reputation of the tags of those users who liked this article. For example, if someone who has never published on the food tag likes Lennet, her article will not receive any rating from this tag.

What does it give us ??

• The rating will satisfy the tastes of the majority, since it is formed by “popularly recognized” platform specialists. Recognition of the value of the article (= Like) by the “popularly recognized” platform specialists actually guarantees high chances for its success with the majority of the community (the reputation of the tags will be higher for those who like the majority, and those in turn will form a rating of articles for the majority - feedback is implemented in the system, in the form of a kind of delegation of tastes ( this is the basis of the idea of ​​rating ).
• The reputation of users is always growing , and is always smeared with the growth of the number of users. Those. It is impossible to cheat your reputation by tags and constantly influence the rating. To make this possible, it is necessary to publish articles that will be liked by as many people as possible, and even by all tags. And with the growing number of people in the system, the contribution from one will be less and less. Those. in such conditions, it becomes almost impossible to manipulate a rating (practically - because I know about probability theory, from the point of view of which I can be a stranger with a probability> 0).
• In view of the point above, we receive immunity not only from cheating, but from cheating by bots ! Bot-no becomes useless in this ecosystem. To write articles that hook human souls - this is not captcha for you to enter)
• The concept of a certain tag "expert". Ladies who publish articles with the “recipe” tag do not affect the reputation of articles with the “synchrophasotron” or “hadron-collider” tags. Those. people who constantly publish "seals" will not ruin the lives of others, all have "circles of interest." )
• Reputation by user tag grows with the platform, i.e. it is a non-constant value, and with a large audience we will not get constancy (dynamics in the form of idol changes).
• In this scenario, not a quantitative, but a qualitative parameter is more important. And this, in turn, also goes to the benefit of the platform itself, providing incentives for the “winning the hearts” of a larger audience .
• And, of course, we are throwing out the self-twisting factor, why should we?

Well, the most interesting: and what in practice?

But in practice we have the following: the rating of articles is published on the golos.io platform every day for many months. According to the feedback of users, the idea of ​​rating is hard to perceive (explanations are not my strong point), but as soon as people begin to understand it, they already perceive it with enthusiasm and as a certain element of gamification. And of course, the rating fulfills its main goal - recommends “tasty” among the articles.



Of course, the platform has not so many curators (= those who like) per day, and only the last weeks have exceeded 1000 articles, which does not allow us to fully appreciate all the charms of the rating (although for the same steemit.com, an English-speaking brother ", Activity is higher and the dynamics in the rating is more high).



Recently, I began to publish in parallel the second version of the rating - for novice bloggers. The idea is the same, with the only difference that the articles of authors with high reputations are eliminated; a kind of rating "sandbox".



And this version of the rating pleased me, because with a small platform, it:

• perfectly displays original articles newbies;
• the rating limit "top" does not allow stagnation in the rating to authors with one style of articles;
• all promising beginners are visible, which is a big plus for the development of the platform.

PS

In general, the idea of ​​the rating is also beautiful because it can be adapted to many areas and get similar results, for example, for search results or reviews in the store, everything depends on your imagination.