Topic: Forum Metrics - Section/subsection sMerit breakdown in detail (Read 514 times)

This confirms what I suspected: that the subforum with the highest signal to noise ratio in the altcoins section, mining, gets a disproportionately low amount of merit. A real shame, since there is very little of the rancid "good project" and "when moon" shitposting that is the hallmark of the announcements and discussion subforums.

As you mentioned Portuguese board,which is my local, I think I know what happened.

One legendary created this topic on 22 April and started to give merits to Good old posts
https://bitcointalk.org/index.php?topic=3370827.0;topicseen

I didn't knew the impact was so big. We are clearly out of smerits there :/

But like the Spanish board, as you said,  the number of good posts per week is small

Clearly it is one of my favourites charts. You can see that the local boards on the overall seem to follow a descending trend until mid-march, then seemed to stabilize until just before mid-april, and then seem to initiate another trip to the marsh lands.
But if you see them on a one per one basis, they have follow different individual patterns. For example, the Portuguese board nearly triple the its trend on the week of the 23rd of April. So did the Philippine board. Russian and Turkish boards spiked up too that week, but German and Italian took a dip.

The thing is, since the amount of weekly sMerit awarded on local boards is relatively small (I’d love to know the “active” size of each of these communities), events can shift things quite a bit from one week to another.

Most local boards are so residual denoting either a total lack of sources or meritable posts (or both). That is where each local board knows its own context, and can bring a spot of light as to the reasons.

For example, the Spanish board is very low in overall sMerits, but the number of new decent posts per week in my opinion is small and comes from a number of users I can count with both hands being generous. The tone in which things are approached is also different to other boards, and I hardly see any newbie activity really (a bit of quick attempts to increase activity exist though).

Of course I will make a reference to your original post. Don't worry.

I agree that the number of posts should be considered, as well with the quality.
I believe all local boards share a lower quality than the main boards. But those boards are clearly being more merited.

This is an interesting one too. Those boards are running out of smerits.

Bitmover, fine by me. Some forum users have asked me similar things for the Turkish and Philippine board (total translation or partial replica). A wee reference to the OP would be appreciated, but not mandatory.

The idea should be to spin dialogue on other boards, since I believe the information shown has a lot for discussion, especially on local boards.
It is important though to consider that data does not say it all. As I mention in the OP:

Ideally, this graph should be contrasted with the number of posts in each section/subsection since sMerit Kickoff (there is more than meets the eye to do this, and it would have some other sub-issues that would need to be considered), and also with a “quality” measure of some kind.
For example, a local language board may seem to be very scarcely merited, but without comparing to number of posts, and out of them those with a good standard, we would be jumping to conclusions just based on one of the variables of a multivariable-based problem.

I think that this graph is very disturbing



Announcements, Russian  and Turkish are clearly unbalanced.

Announcements shouldn't receive all those merits. Spammers are getting merit there and ranking up. This looks like a problem in the merit system. Merit system was designed especially to prevent spammers from ranking up.

Why so many Russian and Turkish merit sources??
This should be better distributed.

@DdmrDdmr, can I post some of these graphics in another sessions? I would like to show on my local board

My view is the same as in my previous post about alts in general, merit was quickly traded, but no more is coming from sources.

 I was in favour of creating a "Serious Discussion" just for Alts and tokens because I think that we, as users, deserve to talk seriously about a whole universe of coins and tokens that are not being discussed due to thousands of one-liners, but it did not have enough support.

It will disappear by itself, since sources are probably not awarding much merit on alts sections.

[ANN] thread is amongst of top merited. Do the merit sources often visit the Announcement board?

Russian and Turkish local forums are ones where merits flow. Is it a coincidence that most of the merit exchangers caught were speaking those languages? Russian dude had even his life ruined because of negative trust.

I believe that that is one of the great questions pending answer related to the Merit System. Each month we’re seeing a decline in the total sMerit being awarded. Maybe our standards are increasing as time goes by, thus awarding less. Perhaps there are less meritable posts over time. Or it could be down to overall total potential awardable sMerit decreasing, possibly at a heavier rate that was expected.

The interesting breakdown would be:
-   Total overall awardable sMerit by Users.
-   Broken down by subtotal and number of users in a categorization such as (users with >0 sMerit):
*         0% awardable over Total personal sMerit
*         Between (0% and 10%] awardable over Total personal sMerit
*         <…>
*         Between (90% and 100%) awardable over Total personal sMerit
*        100% awardable over Total personal sMerit
-   Same for Merit Sources.

Perhaps crossed with a category that indicates the type of awarder one is viewing past history (High awarder, Mid awarder, Low awarder, No awarder).

Since this all evolves, ideally this would be monitored over time. I’m not saying the above should be nominal – a global stat would suffice.

The breakdown would be really good to know, since it would show us just how much potential sMerit there is in the Forum, and of that how much is not really likely to be circulating in practice (i.e. users that have sMerits but have never awarded before ->100% available+No awarder ).

After all, a certain degree of overall supply of sMerit should exist for the system to be effective and working in favour of it’s objectives.
Taken to extremes, if sMerit supply is really high then it becomes useless, but if it extremely low then it wouldn’t incentive posting with good standards at all. The answer is obviously in between, but there should probably be upper and lower tolerance range that optimizes system performance against its objectives.

I disagree, and I'll show you an example why: Merit summary of tonych.
This is exactly what the merit system is designed for. Feel free to read his merited posts, it's a great start to find quality posts on the Altcoin board.

It's a pity that we can't split sMerits from source merits, but I can see that that would be incorrect politically. It would also be nice to know the number of merits available ( source and spendable), and compare that with the number awarded.

It is sad to see that the greatest number of merits are awarded in the alt section. Alt is the scourge of the forum, and I believe that post on those boards should not be meritable.

DdmrDdmr is on fire a lot of good contributions to the community from you you deserve all the merits you have thanks for spending serious time to this forum. Goodluck to your way ranking up. 

It is, rather heavily in Meta. In fact, by looking at the detail in section 5.5 and comparing it to all other 5.x section details, Meta is by far the subsection with the highest quantity and percentage drop between weeks 1 and 2, going from 4.699 sMerits to 1.501 in just a week.

I think Keeping the first week in the data and having it broken down by week is good as it reveals these facts. I would keep it.

As always, amazing data. How difficult would it be to re run the data without the 1st week of the merit system being included, I am pretty sure the data for meta is skewed due to all the initial merit being handed out in here when it was a new system.

This one was a bit of a tough one. I actually had it finished yesterday, but went ahead and refreshed it today to get the latest info available. The editing part is a bit of a pain, but I would say there's at least one relevant item for any one of us here.

Forget the graphical artist bit: I’m color blind, and tend to draw brown pastures grazed by green cows ...

P.D. I updated the sMerit Network today too, but you need to use the new link, since every update makes me redo the support.

I've been following your account since the day you created the looked alike cell about merits. And this post is also quite interesting unlike the posts of those shitty bastards here in the forum. And I guess, you're quite having fun on making charts and graphs that is related here.

I think you should apply on theymos as a GRAPHICAL ARTIST for his posts together with the mods.

Anyways, it is a lovely thing that you're giving us a positive infos like that.

Thats true. It's actually what I ment by the comment in section 4.7.:

Ideally, this graph should be contrasted with the number of posts in each section/subsection since sMerit Kickoff (there is more than meets the eye to do this, and it would have some other sub-issues that would need to be considered), and also with a “quality” measure of some kind.
For example, a local language board may seem to be very scarcely merited, but without comparing to number of posts, and out of them those with a good standard, we would be jumping to conclusions just based on one of the variables of a multivariable-based problem.


Can't get those numbers though on a per week basis, but they would shine some light for sure ..

The only thing that i'm missing is a graph that compares the  amount of merit earned in a section in period X against the amount of posts in the section in period X, and displays it in a "Merit per x amount of posts" kinda way.. ( Not sure if i'm wording it correctly), but what i mean is the following:

Bitcoin discussion has (lets say) atleast x50 the amount of posts Technical Discussion has, thus it would make sense that there was more total merit awarded in that section than in the Technical Discussion.
But relatively Technical discussion might have a higher merit rate p post..

An example of that would be:
50000* posts / 7249 merit vs 1000* posts/ 824 merit

It would give a nice overview where the most "informative/HQ" posts are made.

*Obviously i don't know the number of posts in each section, or if it would be even possible to get data from that.

0. Introduction:
The Merit System has been around for 15 weeks now. The overall sMerit awarded in each section/subsection is a good indicator of what has been going on, but I wanted to go in a bit deeper and see how it evolved over the weeks, so that we could also see it from a time perspective in addition to an aggregate by forum/subforum.
This information has its base in the merit.txt file released today, so it is as up-to-date as can be.

The general idea is to allow us to look at the data and gain insights on sMerit distribution. The conclusion should not be 'post wherever most sMerit is being awarded, since chances are higher'. If that were so, users should hang around Altcoin Ann or learn Russian (spoiler).
This is not a game of placing your horse where there are better odds, but rather breeding your horse to compete at a higher level. Post placement is only a small part of the equation, but the overall idea should be that of bringing something of value to the table, let’s not keep that out of sight.

This analysis shows “what” is happening in terms of sMerit awarding, but cannot go into the “why” it is being done as it is, since many more variables would need to be considered that are not available. Besides, subjectivity plays a fundamental role in sMerit awarding, and that is extremely difficult to analyse.

Warning: The post is long, so do not proceed if you suffer from viewing multiple graphical statistics, or severe headaches due to the presence of tables crowded with crunched numbers.

Note: Numbers are in my local format, so '.' acts as a thousand separator and ',' for decimals.


1. Overall Accumulative per Week
As a reference, I’ve included a weekly overall breakdown of sMerit awarded in the Forum. We know there has been a steep fall overtime, but recent weeks seem to be stabilizing in the 4,5-5K sMerits per week area. Current week is partial, since we’ve still got the weekend to go to complete the week.



2. Overall Accumulative per forum section
The overall amount of sMerit that has been awarded is 145.572. The most awarded section is the cumulative of Local Boards (32,71%) which I will break-down further ahead, being the least awarded section Bitcoin of all (8,93%).
There is an additional 5,16% that has been awarded to posts that were later deleted, adding up to 7.514 sMerits. These are not reverted as we know from user’s merit balance, and not being a great amount overall, the do constitute a fair share of “untraceable” sMerit awarding.




3. Weekly evolution per section
I wanted to see if there was a shift of the share of total sMerit that each Forum Section was receiving over time, and if there was a trend. To follow this up, I’ve broken-down each section by the % of Global sMerit that each main Forum Section has received per week.
Notice that the graphs show the share of overall weekly sMerit assigned to each section, but the table with the data also gives us the weekly detailed aggregates. Tables show a clear weekly descent in sMerit assigned to every section.

I have grouped data by weeks (Monday to Sunday), therefore the last week is incomplete (first week too really, but is was exceptionally high in terms of sMerit awarding) since file is normally generated on Thursdays/Fridays. I've put a different colour on this week’s bar to indicate this fact.

Note: Do not confuse Share of Weekly Total sMerit with the Total amount of sMerit being awarded in a given Forum Section.

3.1 Alternate Cryptocurrencies
This forum section has a slight weekly upward trend in terms of share of sMerit, with a couple of ups and downs, perhaps ICO launching related.



3.2 Bitcoin
This forum section has a rather steep weekly upward trend in terms of share of sMerit, but is rather variable week to week.


3.3 Deleted Posts
Not being a section per-se, there is an acute downward trend of merited posts that get deleted. This could be deceiving though, since it probably takes a few weeks for deletions to occur in its full extension (so most recent weeks will probably be affected over the following weeks).


3.4 Economy
This forum section has a slight weekly upward trend in terms of share of sMerit, with a large variation recently.


3.5 Local
This forum section has a rather steep downward weekly trend in terms of share of sMerit. Full breakdown further ahead.



3.6 Other
This forum section has a slow descending weekly trend in terms of share of sMerit.



4. Overall Accumulative per forum subsection

Forum Sections are made up from by various subsections, so in order to get a clearer vision we need to get a drilled-down view of aggregates per forum subsections to see the composition of a specific forum section.

Each section has three sets of data:
- A summarized table
- A graph (left) that shows the share of sMerit for the Subsection within the Section.
- A graph (right) that shows the share of sMerit for the Subsection in the overall total sMerit aggregate frame.

4.1 Alternate Cryptocurrencies
Announcement (Altcoins) rules here, with a 56,27% share of the Forum Section and is one of the leading overall sMerited sections (13,52% of total Forum sMerit). Mining and Speculation are rather measly in comparison: nearly 10 times less.


4.2 Bitcoin
Bitcoin Discussion is the boss in its Forum Section (55,79%) and sums up 4,98% of the Total Forum sMerit. This should be a nice section to read, but from my personal view, it´s rather full of spam in mega-threads that make it rather cumbersome.


4.3 Deleted Posts
Can’t say too much here, but the aggregate is just above what actually gets awarded to the BitCoin Discussion subsection, so specific weight of deleted posts is noticeable.
We cannot see where the deleted topics were originally placed to get deeper insights here.


4.4 Economy
Economics (46,61%) takes the lead here over Marketplace (38,69%), both being noticeable in the overall Forum Share too.


4.5 Local
Share is extremely fragmented based on which Local board we’re considering.  The Russian board is clearly the leader here (38,11%) and is very important in the overall Forum picture (12,47%).
With 20 local board (and an Other Languages subsection), only Russian, Turkish, Indonesian and German boards have a share above the share mean. The vast majority have a less than 1% Local Board share, and are nearly vanished from the Global Forum sMerit share’s perspective.


4.6 Other
Meta (70,90%) rules in the Other Forum Section, whilst the Beginner’s section trails along in second place with only 11,55%. This is obviously correlated to the profile of the posters in each of these sections.


4.7 Overall Forum/Subforum share
This is one of my favourite views, and it may help discover where there is a lack of Merit Sources. Nevertheless, that depends on other factors that I cannot combine here due to lack of information to cross.

Ideally, this graph should be contrasted with the number of posts in each section/subsection since sMerit Kickoff (there is more than meets the eye to do this, and it would have some other sub-issues that would need to be considered), and also with a “quality” measure of some kind.
For example, a local language board may seem to be very scarcely merited, but without comparing to number of posts, and out of them those with a good standard, we would be jumping to conclusions just based on one of the variables of a multivariable-based problem.



5. Breakdown of sMerit by Subsection
I wanted to include a detailed breakdown of sMerit awarded per Subsection and per week. I did this specifically because I wanted to track the results on my local board, but then saw that this could be useful for general use, so I decided to extend it to all section/subsections.
The general weekly sMerit trend we know is descending, but this help us to figure out where and just how much the descending trend goes on a weekly basis.
I will not place specific comments here, since they should be based on readers section/subsection preferences.

Note: If you’ve got down to these charts, you’ll probably find them to be more interesting than all the above due to the degree of detail on the data tables.

5.1 Alternate Cryptocurrencies


5.2 Bitcoin


5.3 Deleted Posts


5.4 Economy


5.5 Local


5.6 Other