Pages:
Author

Topic: Merit network analysis: merit rank distribution and satellites (Read 1170 times)

hero member
Activity: 536
Merit: 513
Nice graphical representation. Can I know what application you use to do that? I just wondering, will it be more informative if those lines have legends?
There are too many edges and nodes to be able to display more information, as you can see from the images. So the colors and bigger names are one way to show major connections and visualize patterns.
Right, legends would be overlapped and prevent to see the structure of the networks.  Instead one can use weighted arrows like figures in https://bitcointalksearch.org/topic/global-structure-of-merit-networks-3650124

As for the tools, I wrote about it in other thread, https://bitcointalksearch.org/topic/eigenvectors-in-merit-network-space-3803541 let me quote it here:

Finally, several people asked me what kind of tools I am using for these network analyses, and I presume there are more potential interests, so let me answer here.  I am using igraph on python and Gephi, which are tools designed for social network analysis (SNA).  There are many softwares and programming tools for SNA, see e.g. the list in Wikipedia [4], among which you may find your favorite ones.

-snip-

[4] Wikipedia, https://en.wikipedia.org/wiki/Social_network_analysis_software#Types_of_Software
full member
Activity: 294
Merit: 100
Very nice work, I'm particularly excited about what this can do in terms of finding those who trade merit between their alt accounts. I think this highlights something that I've been saying for the longest while, there needs to be a better system to distribute merits, the bulk of the forum has received between 1 and 9 merits since the merit system was introduced months ago. Its clear that there is a need for more merit sources. Deteriorating the sMerit of those who are refusing to send them would be a good way to increase merit distribution.

This merit system has good intentions for the forum but as a general user I find it hard to get the merits. Even on posts which are good I have not recieved merits. I think Merit sources should be more.

Some suggestion from the source point:


1/ Giving merit to the USER who reports USERs who are abusing the forum like sending merits between themselves.
2/ Airdropping the merits from the smerits which depreciated bcz they were not sent to the users>>>It can be distributed based on the ratios or can be random.
hero member
Activity: 784
Merit: 1416
Nice graphical representation. Can I know what application you use to do that? I just wondering, will it be more informative if those lines have legends?

There are too many edges and nodes to be able to display more information, as you can see from the images. So the colors and bigger names are one way to show major connections and visualize patterns.
full member
Activity: 196
Merit: 274
Nice graphical representation. Can I know what application you use to do that? I just wondering, will it be more informative if those lines have legends?
legendary
Activity: 2814
Merit: 2472
https://JetCash.com
Nice analyst report. But seem lower ranker give away merit more than higher ranker. 

That isn't possible. Merit sources are chosen from the higher ranks, and the 'lower' ranks will need to receive their sMerit from them. This means that the higher ranks must award more merit.
newbie
Activity: 163
Merit: 0
Nice analyst report. But seem lower ranker give away merit more than higher ranker. 
hero member
Activity: 536
Merit: 513
Hi, did I post the theme correctly? Or better here?
https://bitcointalksearch.org/topic/m.38607812
The topic seems more suitable for Meta.  You can move it using the "move topic" link at the bottom left of your topic.
member
Activity: 238
Merit: 15
Hi, did I post the theme correctly? Or better here?
https://bitcointalksearch.org/topic/m.38607812
legendary
Activity: 2814
Merit: 2472
https://JetCash.com
We must admit that the system with merits is not successful.
Not only that there are markets, so now there will still be ubiquitous bans, erroneous bans, and other problems. Just now, it's almost impossible to finish the shaking of the hiro. But only of course if you do not clip posts. A simple member of society, it is now not possible, IMHO.

Written using google translit.

I didn't understand your post. You need to discover a different method to create English posts.
sr. member
Activity: 778
Merit: 291
40% is 40%
We must admit that the system with merits is not successful.
Not only that there are markets, so now there will still be ubiquitous bans, erroneous bans, and other problems. Just now, it's almost impossible to finish the shaking of the hiro. But only of course if you do not clip posts. A simple member of society, it is now not possible, IMHO.

Written using google translit.
legendary
Activity: 2310
Merit: 2073
Very interesting material. I would like to transfer merit for cheating accounts were punished by the forum. So how is it that the system of merit becomes more closed because of certain people.
legendary
Activity: 3332
Merit: 3116
Don't forget that almost half of the merits available don't come from merit sources. For every merit a merit source awards, almost another merit is created for the community to award. In theory this should expand the distribution of merits across the forum.

The problem is the merit starvation in some local boards. For instance, in the Spanish board, merits go from one member to another, but they are few, and only the ones who "go out" and write in the English one can receive enough merit to distribute again in their local one.
That's why maybe the Merit Sources should be dedicated and every local might have one at least (depending, of course, on how many members and posting activity has the board).

legendary
Activity: 2814
Merit: 2472
https://JetCash.com
Don't forget that almost half of the merits available don't come from merit sources. For every merit a merit source awards, almost another merit is created for the community to award. In theory this should expand the distribution of merits across the forum.
legendary
Activity: 2338
Merit: 10802
There are lies, damned lies and statistics. MTwain
<...>
The key is determining where the merit sources have to be placed. The who comes later but the mandatory first task is to determine where (section/subsection) and why (numerical justification based on Nº sMerits awarded; Nº posts lacking merits (aprox. human scan in section/subsection)), all in a given timeframe (i.e montly).

Mods as you suggest, or heavy posters/awarders in specific sections/subsections have the topic knowledge that enables them to be better judges when it comes to sMerit awarding).
legendary
Activity: 3332
Merit: 3116


Now what’s left is determining the sorce/sources. That is where we would postulate as candidates users that award most in the given section/subsection, combined with a wider spread of awarding history over multiple posts and/or users. That is, better to have as a source someone who awards few sMerits to many users that a lot to a few users.
Divide the sMerit deficit by the average you want to give to a source, and you get the number of sourced to have in a given section/subsection.

It’s not perfect as finer details need to be outlined, but that’s how I would approach it and build up from there.

There’s one more thing that buggers me which is, regardless of it all, what is a good overall sMerit circulation that is needed to keep the system healthy, working, and having users believe in it?.
Too much sMerit circulation would render it useless, as it would defeat the purpose, but having a very restricted amount may drown the system and it’s credibility through inanition.


Even when it seems difficult to be done, that's a great idea. I just see a problem in detecting "good posts" in between such a daily amount of newcomers and new posters...
So, based on your idea, another one comes to me. Maybe merit sources can be chosen per area. I mean: not everyone loves to discuss society, and not everyone loves to discuss tech, for instance. So let us select the merit sources per area: for instance, Flying Hellfish can be the merit source in the Politics&Society area, and he can be dedicated to merit this section.
But, how many merits will the area receive? It just depends on how much activity the area has. If we select for instance a local board, we can make an analysis of how many users are in there, how many posts per month they are used to have, how much are they growing up, how many mods do they have... and etc. So the Smerit can be given to the local mods (or other useful users in the area), and they will share it in their local board only.
The meta section, again, the same. How many posts do we have per month? how many users are active in this section? how much is it growing? after an analysis of each section, we can determinate (approximately, of course), how many Smerit a section will need.

Of course, it is impossible for it to be constant. We can have a nice week, with many nice posts, and a bad one with none to be merited. But the common situation will be, at least, a couple of good posts per day (hopefully!!!). Being in communication with the merit sources in each area can be really refreshing to the system.
I also think this is necessary to have global merit Sources in order to refresh some areas, but it can be a way to refresh the merit system and to encourage people to make some good posts.

Just an idea...


member
Activity: 238
Merit: 10
sMerit does not resume and it becomes harder and harder to get them. I met on the forum participants who are doing a very big and useful work, and do not pay for it sMerit. What will happen next when all the plugs are destroyed? Or once a year sMerit will be renewed?
hero member
Activity: 784
Merit: 1416
The board id information would be redundant as is not adding anything new, you have already the message id, from that you can get that information, i have done it already( https://bitcointalksearch.org/topic/m.34584678)
hero member
Activity: 536
Merit: 513
Very nice work, I'm particularly excited about what this can do in terms of finding those who trade merit between their alt accounts.
When I saw this:

from your user name I expected some negative feedback but good to hear my point is valid Smiley



These are very useful analyses, great job!

I am particularly interested in identifying people who should be made merit sources, or existing merit sources who should be given a higher monthly limit. When I originally created the system, I thought it'd be fairly easy to look through the stats and figure out who would make good sources, but it's been more difficult than I expected. Is there any additional data I can provide that would be useful?
As I requested here, I was interested in having board ID in data file.  It would be useful if you could add it as follows:  

The current format:
Code:
time amount msg user_from user_to
1516831941 1 2818066.msg28853325 35 877396

Add the board ID:
Code:
time amount board msg user_from user_to
1516831941 1 24 2818066.msg28853325 35 877396

as it would allow us to analyze merit circulation at board level, as DdmrDdmr also wrote above.



How are "mutual merit transactions" weighed in this model, such that they would appear as isolated satellites? For instance, Coin++ and CarlOrff appear completely isolated. But if you look at CarlOrff's merit summary, he has sent merit to/received merit from other people besides Coin++ (albeit less).
The small figures in Sec. 2 are cropped from "Full Member" network, so other merit rank accounts are not shown.  Coin++ and CarlOrff have merit transactions to other accounts, which are not in "Full Member" merit rank.



There’s one more thing that buggers me which is, regardless of it all, what is a good overall sMerit circulation that is needed to keep the system healthy, working, and having users believe in it?.
Too much sMerit circulation would render it useless, as it would defeat the purpose, but having a very restricted amount may drown the system and it’s credibility through inanition.
I am also interested in hearing what theymos has in mind about this.  Intuitively, at least up to "Member" merit rank the merit circulation looks like sufficient, though quantitative analysis would be required to make more precise statement.
legendary
Activity: 2338
Merit: 10802
There are lies, damned lies and statistics. MTwain
<...>
I am particularly interested in identifying people who should be made merit sources, or existing merit sources who should be given a higher monthly limit. When I originally created the system, I thought it'd be fairly easy to look through the stats and figure out who would make good sources, but it's been more difficult than I expected. Is there any additional data I can provide that would be useful?

In order to help identify merit sources, this kind of analysis is a step towards it, but there are a few things to consider in my opinion:

1. The first need is to identify merit awarding lacking areas. A network analysis of already existing sMerit awarding TXs, being it either from o global perspective or a local one as was my approach, is a path to identifying merit abuse and possibly alt account ultimately. Nevertheless, it is based on user to user relations as a primal driver, where posts would need to be the focus to identify Merit awarding lacks.

My approach would be to analyse boards, at the very least at section/subsection level. For each of these we have to assert if there is a need for additional merit sources. This analysis is threefold:
a) Determination of the current amount of sMerit and Users awarded in the section/subsection.
b) Determination of the number of posts in the section/subsection.
c) “Quantification” of the quality of the posts being posted in the section/subsection.   

These are not simple isolated dimensions, since they should be measured in a certain timeframe. The overall measure of sMerit awarded is fine for a general overview, but in order to asses if a new merit source is needed we should limit it to a given timeframe.
Just as to determine the stock of raw material needed for production we need to see it in a narrow framework of time and not considering all the historical data for day to day operations.

“a” is covered and can be traced I guess internally. We got a pretty good glimpse of it last week ( Forum Metrics - Section/subsection sMerit breakdown ->section 5), where we got to see a weekly breakdown of awarded sMerit per section/subsection.
We should forget the first two or three weeks of sMerit fiesta and focus on the rest. That is the real week to week sMerit awarding that takes place in a given section/subsection. Let’s consider sMerit the boxes to contain the produced goods.

“b” is the actual production (number of posts in the section/subsection). We know that posts are awarded retrospectively, but the vast majority are awarded within a week or two timeframe at most from the moment it was created. Alas, we don´t have those figures. Perhaps the number of distinct associated users involved would also be useful.
“b” is not considered in a network approach sMerited TXs, since the view is biased to merited posts specifically, not total number of posts.

“c” is the human quality control of production that must be performed upon “b”. The idea is to give us a quantitate measure of the qualitive side to the posts. This is the toughest one to get right, and should yield as much objective criteria as possible.

Let’s take for example the Spanish Local Board. It was awarded 35 sMerits on the week of the 23/07/2018. Let´s say that there were 350 posts created that week , and that the overall perception is that those 500 posts had an overall quality translated into a “15/100” (that is, the perception is that 15% should be awarded sMerits).
The above scenario would play out to say that roughly:
500 * 15% = 75 sMerits is the minimum expected awarding for this weeks batch. Since only 35 were given out, it would mean that there is a deficit of another 40 sMerits to apply to normalize reality to expectancy, thus a source would be required.

I know, the above is rather simply played out, and the hard part is getting the “c” right since it is key in the dimensioning process. There’s also the fact that some posts should be awarded multiple sMerits, but the principle is still bases on these three factors. It can be further complicated by bringing the number of posters into the equation too.
The above can be set with a weekly timeframe in mind, or something larger like a monthly window.

Now what’s left is determining the sorce/sources. That is where we would postulate as candidates users that award most in the given section/subsection, combined with a wider spread of awarding history over multiple posts and/or users. That is, better to have as a source someone who awards few sMerits to many users that a lot to a few users.
Divide the sMerit deficit by the average you want to give to a source, and you get the number of sourced to have in a given section/subsection.

It’s not perfect as finer details need to be outlined, but that’s how I would approach it and build up from there.

There’s one more thing that buggers me which is, regardless of it all, what is a good overall sMerit circulation that is needed to keep the system healthy, working, and having users believe in it?.
Too much sMerit circulation would render it useless, as it would defeat the purpose, but having a very restricted amount may drown the system and it’s credibility through inanition.
hero member
Activity: 1022
Merit: 564
Need some spare btc for a new PC

How? That will just decrease the available supply of sMerit. Unless you're talking about redistributing their sMerit to other people?


if you ask me, someone holding on to their sMerit is somewhat the same as not giving them away. No ones profiting from that, so maybe they should be deteriorating, after a while.
Pages:
Jump to: