Ratings of people: a new hero of a future era

! The current state of the topic is now supported in this Google document .
People rating services can become the controlling link in the link of brands - sites - users, change the balance of power in the existing ecosystem of various sites, and also introduce new opportunities for network social interactions. Moreover, they are able to extend their influence to offline, to the relationship of businesses and customers or customers with customers. Under the cut a lot of letters in support of these theses and other related topics.

Positioning

To begin with, we note the importance of accurately determining what exactly is being measured, which is the higher / lower criterion for a person’s position in the ranking. For a task in such a general form as stated above, the discussion should probably be about the level of authority and influence in society. I prefer to use the term "social weight". Of course, this is a difficult question - influence can be given by a position, influence can be given by money. But taking a position and having money, you can be a completely worthless non-authoritative person. Or by cheap pop music you can influence the army of fans. I will omit these problems now, because I think for most people this is not so relevant. We will dig too deeply and move away from the stated goal if we start now to take it all apart. I think the term “social weight” is quite intuitive and adequately describes what is needed.

The existing and already pretty popular Klout project also measures influence, but not in society at all, but in some of the most popular social networks. Say, there is a certain correlation between offline and Internet influence - people with a high Klout rating are often (although not always) quite successful offline. The problem is that the opposite is not true - reputable offline people do not always have a high Klout rating. It is clear that the easiest way is to read the information from several large social networks, but in this way the problem in the complex cannot be solved. Or to solve only part of the problems, which is also not bad.

The second important point related to positioning. For most people, the main channel through which their social self-realization goes, gaining authority, influence and weight is a professional activity. The question "who is this person?" Is usually called a profession. Therefore, it is tempting to partially equate the social weight and value of a person as a specialist in a certain field. Naturally, these are related things, but they cannot be mixed in one rating service. Rating in the professional sphere has its own specifics, other algorithms, such a service has a different target audience and may have a different monetization, as will be shown below. By claiming that the Klout rating can influence your career, Klout tries to mix these things. Such a policy (albeit unobtrusive) harms Klout, because, without being imprisoned for this task,

Rating - brands - sites - users

We are now considering the rating by social weight, not by profession. The connection of the ranking with the brands is probably a find of Klout and its huge merit. Like Groupon, he formed a new discount model: the higher the user rating, the more bonuses and perks are received from brands, businesses offline and on the Internet. Especially if the user has any thematic specialization. The logic is this - an influential user who has received a discount from the brand becomes loyal to the brand and maybe even a little of his evangelist. At least, he will probably write on his blog that he bought an iPhone for a third of the price (or received it for nothing) and all five thousand loyal readers for whom he is credible will thereby receive their share of free Apple advertising. I came across examples showing that such a strategy can be many times more effective than traditional advertising. There is also a game of elitism - after all, not everyone can get a discount. Moreover, according to the principle “they don’t look in the teeth for a gift horse”, the recipient of the discount has psychologically less incentive to criticize the product.

Klout is now actively kneading this market, bringing these ideas to brands and users. And it seems more to brands, because for users on the Klout website, the Perks section is not striking, some do not even suspect the possibility of converting their rating into real benefits. Here, the story is somewhat similar to Groupon - discounts and bonuses are often a regionally attached thing, for example, now Klout-perks operate mainly in the United States. What is surprising to me is that Groupon spawned a whole explosion of clones all over the world, which apparently forced him to spend large sums of received investments on the quick purchase of regional clones. And in the case of Klout, nothing of the kind is observed. This service not only does not work with Russian brands, but also does not process data from popular Russian social networks, however, I have not seen any noticeable clones in Russia. Apparently, this topic is specific to understanding by both start-ups and investors; I hope this text will still serve to attract attention and make sense.

Thus,
1. users are interested in a rating service in order to increase their rating in order to receive benefits from brands.
2. brands are interested in rating service for the sake of effective promotion.
3. Since various sites are the main place where users can increase their rating, connecting sites to the rating system stimulates user activity on them. Therefore, sites are also interested in a rating service.

None of the links in this chain can normally interact directly with others, it can only through an intermediary - a rating service. Therefore, the more popular the rating, the more powerful it is, it begins to dictate terms. We also note that there cannot be many rating services by their very nature. They reduce a huge variety of data to one number; as my friend says, simplify the task to cowards. A large number of rating services would nullify this advantage. Exactly for the same reason, there are not many search services in the world - after all, these are the same ranked lists, only for sites. By this analogy, we can predict the future - most likely someone will dominate all over the world alone, as in the current situation with Google and the rest. Now those regional and national players who cut the chip, there is a chance to seize the local market and subsequently long and successfully resist the global monopoly, as Yandex in Russia and Sezny in the Czech Republic managed to do. Or become global monopolists ourselves. We can also look at the experience of rivalry between former search favorite Yahu and the then newcomer Google - not the fact that today's favorite Klout will become a monopolist in the future. As I showed above and I will show below, Klout is not all perfect and can be circumvented if it is sensible to play on its shortcomings. We can also look at the experience of rivalry between former search favorite Yahu and the then newcomer Google - not the fact that today's favorite Klout will become a monopolist in the future. As I showed above and I will show below, Klout is not all perfect and can be circumvented if it is sensible to play on its shortcomings. We can also look at the experience of rivalry between former search favorite Yahu and the then newcomer Google - not the fact that today's favorite Klout will become a monopolist in the future. As I showed above and I will show below, Klout is not all perfect and can be circumvented if it is sensible to play on its shortcomings.

I’ll add a little more to paragraph 3, because brand bonuses are not the only incentive for users. Even without any bonuses, the incentive is all the same: the user does not need to earn a reputation from scratch on every new site. The rating becomes "transportable", for example, an IT professional with a high rating will no longer need to go through the Habra sandbox. Reputation is something that is earned gradually, including achievements offline, as it is in real life, and does not change quickly and much depending on successfully shared or uploaded posts, as is the case in Klout (well, or recently because they regularly change something in the algorithm). However, such quick-changeability in one case turned out to be useful, it turned out to be an experiment confirming my initial idea that a site not connected to the rating would lose its users and lose to competitors. When Google+ appeared and was not processed by Klout, for a month of activity in Google Plus, users noticeably lost in terms of their Klout rating, which prompted them to return to Facebook. I do not claim that this is the main reason for the outflow from Google Plus, just such a phenomenon was observed. Which also remotely looks like if some site were not indexed by search engines right now.
Well, we will not forget one more thing - although the rating according to the criterion of professionalism should be considered as separate, it is not completely independent, the component of social weight should probably be present there. As the opposite is naturally true, when calculating social weight, professional achievements must be taken into account. So in connection with professional rating, there is still a whole set of all kinds of stimulants, as discussed below.

Rating as a B2B service The

controlling position of the rating in the chain of brands - sites - users is certainly the most attractive aspect in the strategic plan. But a social weight rating may have “usual” uses:

1. With various kinds of interactions of strangers, the rating helps to make a choice in a variety of situations of choice. For example, on sites like Airbnb, when you rent an apartment with strangers or rent them out to strangers, the rating will help you decide, especially in cases where you are hesitant to choose. The same (even more urgent) in the services of automobile companions. Also, collective loans to each other, auctions, etc.
2. The quality of the content produced by him correlates with the “quality” of a person. This idea in some of its variants is propagated by Witology and a whole set of applications is based on it:
- screenings of bots and trolls in the comments on various media resources, in social networks, blogs. Just turn on the filter by rating level, bots and trolls must have a low rating (by the way, this is one of the tests for rating quality). This idea was expressed by Ivan Begtin back in 2008.
- spam elimination in mail services. Something, although not quite similar, is already being tried to be implemented .
- the opposite option - collecting content "cream" from high-ranking users on different resources, on different topics. In other words, the automatic aggregation of quality content from the network.
- You can also think of a new approach to creating elite communities in terms of rating. Now we have practically only two options - according to the recommendations or the “money filter”, as in Snob.
- Using the rating in crowdsourcing projects to identify and involve potentially the best participants in them. This is one of the important areas, as crowdsourcing itself is a global trend.
- You can think of different services that are either there, but imperfect, or they are not, but with the advent of the rating they may appear. For example, improving friend recommendations . Or in geo-services search in the immediate vicinity of highly rated people from the user's interests.

Calculation algorithm

It would seem that the thing in this text is secondary and optional. But there are some nuances. It is clear that over time, any algorithm will begin to become more complex, include new factors, and with the increase in the number of users, the amount of processed information will dramatically increase. If you do not initially choose a competent approach, you will have to redo it in the process, which for users is fraught with jumps in the rating, then to plus or minus, and these jumps will not be connected in any way with the behavior of users and the actual change in their reputation. This phenomenon is completely undesirable, because demoralizes users. As far as I can draw conclusions from observations of Klout, he has this problem and has not solved it. Moreover, it’s not just technical difficulties, the more serious problem lies in the vagueness of Klout’s positioning - it’s as if trying himself in different ways, and probably changes the algorithm accordingly. But it’s very difficult to improve something if you don’t understand clearly what exactly you are improving, where to move, what are the better / worse criteria. This problem may be exacerbated by the growing popularity of Klout.

My suggestion in this regard is to generalize the already tested well-known PageRank algorithm, which allows you to take into account both quantitative and qualitative indicators and is optimized for large volumes of data from a computational point of view. I propose to generalize the concept of links in the PageRank method: any portion of attention given to someone by someone is a “voting link”. Moreover, the forms of manifestation of attention can be different - comments, likes, sharing, links, quotes, adding to favorites, just reading, fouling, etc. Such a generalization must be done carefully from the point of view of mathematics, but it does not seem to have fundamental problems. Moreover, there are considerations, that the PageRank method can be good also on the other, unexpected side - in it the magnitude of the factors affecting the rating does not need to be set by hand, because these factors themselves are also ranked.

Another important consideration on the topic - in fact, in order to rank a person, you need to rank a lot of different entities related to him. Apparently, the most influential types of entities are authoring content and organizations in which the specialist worked (or projects in which he participated). This is a natural thing, for example, here on Habré in the ranking of a person, the successes of the posts and comments published by him are taken into account. In the proposed generalized PageRank method, it’s actually not people who are rated, but any objects that belong to a common network, including people. Those. in fact, we simultaneously have, for example, also a rating of organizations and content. It is clear that this approach has great potential.

Rating of specialists

It seems to me something very similar to Google’s search page - you enter a set of keywords such as city, programmer, language, platform, framework and get a ranked list of people matching these criteria.

Who needs it:

1a. Direct employers to find the best professionals. It is clear that we are talking more about rough preliminary selection, but this already saves a lot of time and effort in the search.
1b. To recruitment agencies for the same purpose. Monetization: fee for advanced information on candidates. After all, the rating is built on the basis of processing at least the same as what is in the questionnaires on resources such as HeadHunter.ru Ideally, there should certainly be more information. In particular, the information that was used in calculating the rating by social weight.
2. Hired specialists need a high rating as an argument for salary and career growth. Well, a low rating of employees will be useful to employers as an argument not to raise salaries :)
3. For specialists who provide any paid services. Tutors, lawyers, psychologists, realtors, car mechanics, etc., etc. Freelancers naturally too. This is a separate song. Monetization: providing a specialist with a personal account on the site and charging a proportional attendance fee for this account. The logic here is simple: the stream of specialists viewing the page is converted to its clients. The minimum functionality of the cabinet is a regular profile, just to establish contact between the client and the specialist.
4. Of course, consumers of specialist services need a rating too. If the doctor works in the hospital, you can choose the best doctor. Monetization for them can be advertising.
5. In terms of B2B, a rating service can provide a specialist rating service for third-party sites. For example, tutoring sites where there are either no ratings at all, or they are primitive and vulnerable to cheating.

In terms of monetization, we add, by analogy with search services, a couple of non-rated paid places in the top of issuance, as well as by analogy with freelance sites, a secure transaction service.

Calculation algorithm

Information for the rating by social weight is not collected in one place of the network and, moreover, is not systematized. Information on specialists is usually collected and systematized in the form of questionnaires on sites such as HeadHunter. And it contains exactly what employers are most interested in. It would be strange not to take these differences into account in the rating calculation algorithm. I think instead of PageRank, another method is more suitable here. In a primitive version, it looks like this: if you have several parameters from the questionnaire, such as age and work experience, you rank the questionnaire according to the criterion so that the sum of these parameters is minimal. (More precisely, experience is just the more the better, but for such cases the opposite value can be minimized). Further, the model can be complicated, enter significance factors for various factors and add the factors themselves.

Competitors

For a rating on social weight I do not even see competitors in Russia. The world is an obvious competitor to Klout. But he is not a competitor in all respects, because measuring social weight and measuring influence in social networks are not the same thing. Usually, even if there are no direct competitors, there are those who lose something. In the case of the rating, it seems to me that everyone only wins - because the rating service does not generate its own content, it does not attract anyone’s users to itself. Only advertising budgets from brands are redistributed, yes. And someone loses control: Facebook may think that it controls the social Internet, but it hardly assumes that Facebook itself can be implicitly controlled by a third-party service. A full competitor of a rating service can only be another rating service.

However, this does not apply to the rating of specialists. Everything is classic here - from its development, money and users will lose work sites such as HeadHunter and lose freelance sites. In general, I think that the rating of specialists is a more adequate alternative to freelance exchanges. First of all, because his monetization model does not affect the quality of the rating. And freelance exchanges essentially arrange an auction to sell paid places in their rating. Therefore, their rating ceases to fulfill its main function - to identify objectively the best. And this function is being replaced for customers by another paid service - for money (considerable) special people will find you an artist. As a result, the customer bears a pointless additional expense, and for normal freelancers the service is hellish hell. In short, this is garbage, not a service. The power of the Free-lance project. ru only in its size and in the temporary absence of an alternative, because other freelance sites do not offer anything fundamentally different. The situation is also regrettable that in general the growth of freelance in Russia is a steady trend and there may be a false impression that the Free-lance.ru website is doing everything right, as the number of its users is constantly growing.

If we are not talking about web freelancers, but specialists who provide services offline, this area is completely undeveloped and is waiting in the wings.

In conclusion, I want to note just the connection of both ratings with offline. If a rating measures social weight, including on offline achievements, then in a combination of rating - brands - sites - users, the link "sites" is not fundamentally necessary. Everything that was said above for the Internet will work offline. Theoretically, you can imagine for example a hotel that gives discounts for a rating that is entirely earned offline. Or not providing services to clients with a low rating, for example, in order to create a special atmosphere of elitism or to filter out solvent, but unpleasant customers, spoiling the rest of other vacationers. Another example: the impact of the rating on the issuance of loans by banks.

In this sense, rating can become an instrument of new social segregation. It sounds intimidating, but in fact this instrument can serve positive purposes and create a world in which power, wealth and pop popularity will not be considered the dominant factors in determining a person’s social weight. The direct influence of the rating on the political situation can be thought of if, for example, the electoral system is based not on the equality of votes, but on the social weight of the voter. Or such an aspect of globalization: people with a high rating have visa-free entry to any country.

PS I am open for cooperation; if someone is interested in changing the world with the help of ratings and you are an intelligent programmer who knows a lot about how to read information from the network, why not cooperate? The above, however, is enough to do something without me. Anything can be fast-cooked quickly, and maybe it will even work when there is fish on fishlessness and cancer. And in a normal product, someone has to deal with the algorithm, because general ideas and the final implementation are two different things. Moreover, it is not final, because everything changes and improves, you constantly need to finish the software part, and the design, and clarify the positioning of the service, and the business model ... In short, this is a classic adventure :)