Chronicle of diving habrasila, script rating, attention vampirism and the 25th day of judgment
The story is devoted to the study of the function of user ratings in one social network. How behaves on Habré rating (habrasila)? In issuing the search there is no detailed description of this process, so the discovery is supposed to be called the “law of the 25th Judgment Day”. It was after this period of time, on the 26th day after the last publication, that the rating began to catastrophically fall, striving for the numerical index of karma.
The publication is intended for users interested in natural sciences. Annotation : as a result of a fundamental difference in the average growth of rating and karma indicators during publications for 25 days, the rating experiences a weak attraction to the level of karma, which then shows a one-time collapse, a period of catastrophic fall, and only after 29 days does the law begin to look like back proportional, natural for linear attraction processes according to the 1 / x law. The user's desire to maintain the rating at a distance from the value of karma forces him to constantly work on publications, attracting attention (attention vampirism), which are energetically advantageous to produce more often than once every 25 days.
Long ago, when the earth was small, and users lived in caves, no one knew who controls the rating. We now know that the script does this, but did not know how it does it. The study, which everyone can repeat with the help of Karmagraf (a script that tracks karma and rating through the Habr API ) or manually, sheds light on the nature of the phenomenon.
For the studied data, we will choose the movement of my rating within a month after the last publication (April 1). Until that time, publications were quite frequent — ideal conditions for comparing the “before” and “after” behavior. The balance of benefits for publishing (easy to calculate in the history of the blog) was 253 (134 minus among them). Karma by that time was about 65, i.e. There were 4 times more advantages to the rating. There was also a positive balance of advantages for comments, which can also be counted by walking through the history of your comments, but we will not do this, this is not the point. For 1.5 months, the final rating turned out to be about 185, which is 3 times more karma. Clearly, this is a consequence of the very law of gravity rating to karma, which was once given to the public, and it acted against the background of frequent publications, decreasing moderately, allowing you to maintain a fairly strong imbalance, away from the equilibrium state.
')
Another interesting observation: formally, I could be in the ranking of the first hundred visitors, because among 60-x - 90-s visitors there are many people with a smaller ratio of numbers (for example, 32-92, 70-70, 64-58), but the “personal card” page showed that the script considered me 130th - 140th. That is, the rating of people in the table is considered differently that the rating of the user in individual data. It seems to be something more time-averaged (almost all visitors from the table are more than a year on the site). (Or else, these are specially chosen people.)
But back to the rating. (Over time, I did not record the data and did not use the pocket meter, so the data is written from memory, but accurately reflects its behavior.)
Date 1.04 - 04/04/2010 - rating 188 - 185 (approximately 130th in the rating of habra people).
In the subsequent period, the rating slowly decreased by 1-3 points per day; increased by 0.3-0.4 each time a comment is added. Amazing happened in one day, more precisely, in one night. Rating has a gap. Most likely, after the expiration of 25 days after the last publication, he once dropped about 40-45 units. (There were no other active actions, so the reason is seen in the 25th day.) At the same time, the general rating begins to fall.
April 25 - 166 (approximately 150th in the rating of habra people)
26.04 - 121 (collapse)
27.04 - 110
April 28 - 99 (peak at 10 per day)
April 29 - 90 (630th in the rating of habra people - the position, of course, is not inversely proportional to the rating)
April 30 - 85 (gradual withdrawal from the corkscrew, 5 per day)
1.05 - 80
3.05 - 77 (even smoother descent)
For clarity, the graph is presented in Excel:
It is easy to identify several significant areas of behavior rating:
* Plateau - "fading charm",
* The collapse - "loss of confidence",
* Smooth decline after shock.
The formula "1 / x" is named very tentatively - the data do not allow us to speak about the exact formula of the law. Maybe 1 / x ** 2 or something else. But it is clear that attraction is not linear.
If you received a rating higher than karma, you have become a "vampire of attention." Now, to keep the rating in such an unnatural ratio, you need to regularly feed the attention of readers. At the same time, the pluses for comments will help little and will be able to weakly hold the falling rating for a short time (up to 25 days). Then, the function of clearly unnatural origin begins to work - with chopped collapses and slopes. Perhaps after the next series of publications, you will be able to return to vampirism (attention) again.
Of course, there is the opposite case, when not a rating forces a visitor to publish, and a publication causes a rating to grow. The causal relationship between these phenomena is still poorly understood, as well as the reasons that encourage the user to publish something.
(They are derived from the behavior of the Habr script at the time of observation; it may change in the future. It is likely that the script still works according to the rules summarized here: " Habrasila scoring and rating (11/26/2008).")
If you want to keep high positions in the rating - do not interrupt the publication for a time longer than 25 days - the hopeful script will disappoint you. Publications, probably, should be interesting, so that the collapse does not happen, but I have no data about it. In any case, uninteresting articles will not help either people or rating. And also (by reference to the source), “ The author will not be able to increase the posts on non-core blogs of habrasila. Moreover, if the author does not post anything for a long time, his habrasil will gradually approach the value of karma. Habrasila itself is calculated based on the votes for topics and the author's comments. Moreover, given the complexity of the publication. "( Shoohurt ).
The integral rating of habra people works according to more time-averaged laws than a personal rating (called habrazila). It is by the integral rating that you can see the depth of the script’s frustration in you after the 25th day. If the personal rating fell 1.8 times, then the position among other users - in 4 (due to the exponential law of the distribution of the elect).
The publication is intended for users interested in natural sciences. Annotation : as a result of a fundamental difference in the average growth of rating and karma indicators during publications for 25 days, the rating experiences a weak attraction to the level of karma, which then shows a one-time collapse, a period of catastrophic fall, and only after 29 days does the law begin to look like back proportional, natural for linear attraction processes according to the 1 / x law. The user's desire to maintain the rating at a distance from the value of karma forces him to constantly work on publications, attracting attention (attention vampirism), which are energetically advantageous to produce more often than once every 25 days.
Long ago, when the earth was small, and users lived in caves, no one knew who controls the rating. We now know that the script does this, but did not know how it does it. The study, which everyone can repeat with the help of Karmagraf (a script that tracks karma and rating through the Habr API ) or manually, sheds light on the nature of the phenomenon.
For the studied data, we will choose the movement of my rating within a month after the last publication (April 1). Until that time, publications were quite frequent — ideal conditions for comparing the “before” and “after” behavior. The balance of benefits for publishing (easy to calculate in the history of the blog) was 253 (134 minus among them). Karma by that time was about 65, i.e. There were 4 times more advantages to the rating. There was also a positive balance of advantages for comments, which can also be counted by walking through the history of your comments, but we will not do this, this is not the point. For 1.5 months, the final rating turned out to be about 185, which is 3 times more karma. Clearly, this is a consequence of the very law of gravity rating to karma, which was once given to the public, and it acted against the background of frequent publications, decreasing moderately, allowing you to maintain a fairly strong imbalance, away from the equilibrium state.
')
Another interesting observation: formally, I could be in the ranking of the first hundred visitors, because among 60-x - 90-s visitors there are many people with a smaller ratio of numbers (for example, 32-92, 70-70, 64-58), but the “personal card” page showed that the script considered me 130th - 140th. That is, the rating of people in the table is considered differently that the rating of the user in individual data. It seems to be something more time-averaged (almost all visitors from the table are more than a year on the site). (Or else, these are specially chosen people.)
Value "date - rating" and its movement over time
But back to the rating. (Over time, I did not record the data and did not use the pocket meter, so the data is written from memory, but accurately reflects its behavior.)
Date 1.04 - 04/04/2010 - rating 188 - 185 (approximately 130th in the rating of habra people).
In the subsequent period, the rating slowly decreased by 1-3 points per day; increased by 0.3-0.4 each time a comment is added. Amazing happened in one day, more precisely, in one night. Rating has a gap. Most likely, after the expiration of 25 days after the last publication, he once dropped about 40-45 units. (There were no other active actions, so the reason is seen in the 25th day.) At the same time, the general rating begins to fall.
April 25 - 166 (approximately 150th in the rating of habra people)
26.04 - 121 (collapse)
27.04 - 110
April 28 - 99 (peak at 10 per day)
April 29 - 90 (630th in the rating of habra people - the position, of course, is not inversely proportional to the rating)
April 30 - 85 (gradual withdrawal from the corkscrew, 5 per day)
1.05 - 80
3.05 - 77 (even smoother descent)
For clarity, the graph is presented in Excel:
 |
|---|
* Plateau - "fading charm",
* The collapse - "loss of confidence",
* Smooth decline after shock.
The formula "1 / x" is named very tentatively - the data do not allow us to speak about the exact formula of the law. Maybe 1 / x ** 2 or something else. But it is clear that attraction is not linear.
Vampirism attention people
If you received a rating higher than karma, you have become a "vampire of attention." Now, to keep the rating in such an unnatural ratio, you need to regularly feed the attention of readers. At the same time, the pluses for comments will help little and will be able to weakly hold the falling rating for a short time (up to 25 days). Then, the function of clearly unnatural origin begins to work - with chopped collapses and slopes. Perhaps after the next series of publications, you will be able to return to vampirism (attention) again.
Of course, there is the opposite case, when not a rating forces a visitor to publish, and a publication causes a rating to grow. The causal relationship between these phenomena is still poorly understood, as well as the reasons that encourage the user to publish something.
findings
(They are derived from the behavior of the Habr script at the time of observation; it may change in the future. It is likely that the script still works according to the rules summarized here: " Habrasila scoring and rating (11/26/2008).")
If you want to keep high positions in the rating - do not interrupt the publication for a time longer than 25 days - the hopeful script will disappoint you. Publications, probably, should be interesting, so that the collapse does not happen, but I have no data about it. In any case, uninteresting articles will not help either people or rating. And also (by reference to the source), “ The author will not be able to increase the posts on non-core blogs of habrasila. Moreover, if the author does not post anything for a long time, his habrasil will gradually approach the value of karma. Habrasila itself is calculated based on the votes for topics and the author's comments. Moreover, given the complexity of the publication. "( Shoohurt ).
The integral rating of habra people works according to more time-averaged laws than a personal rating (called habrazila). It is by the integral rating that you can see the depth of the script’s frustration in you after the 25th day. If the personal rating fell 1.8 times, then the position among other users - in 4 (due to the exponential law of the distribution of the elect).
Source: https://habr.com/ru/post/92866/
All Articles