Topic: Analysis- sMerits per transaction and sMerit transactions per post-Are they low? (Read 411 times)

Fair enough, it seems difficult to manage and also pointless, I guess, for the result is something we all know well: most of the posters are shitposters, so they are not just receiving merits.
I just was thinking about this week discussion (how can we approach to the merit system and be "fairer" by helping people to rank-up). After all those insights we have had, I truly think this is going to be a matter of changing mindsets, but also, embracing and accepting the truth: the forum is going to become a place in with there are different kinds of people:
- Those who are here with the only goal or making some bucks = quantity vs quality.
- Those who are in here earning some bucks (why not) but also truly concerned about the forum itself and its functions, engaging good discussions and creating useful content =quality vs quantity.

I don't think it to be a big issue, though. It is how society works in large cities.

Yes, pretty low rate of users that get merited every week (even if we leave margin of missing out a few weeks for any number of reasons). Just goes to show that it is difficult for everyone not just the lower ranks.


Regarding the further analysis proposal, the key would be in being able to play around with a member’s posting history. Now the issue here is that information available to us (outside the Forum’s internal analytical structures) has different levels of granularity and timestamp information:

•   sMerit TXs have a date/time, a message Id,  a userTo, a UserFrom, and the amount of sMerits awarded. So we have knowledge of exactly who awarded how many sMerits to whom, and when it occurred for what message (id).

•   User profile information has only “one” reference timestamp, which is the moment the data is actually scraped. So all the variables related to the profile are relative to a single timestamp. That is, there is no dynamic view on the data with a timestamp reference to when each little detail occurred; we only have a snapshot photograph.


So for example, we can get to know that a certain user has posted 223 posts, but in order to get the 223 post’s header information (messageId, date/time, section/subsection, etc.) you would need to download all the posting history for all the members that one would want to study.
This is unfortunately a very difficult task, since it would mean creating a process that goes into a user profile, goes into his posting history, goes into each post, retrieves each post’s data by scraping, etc.

While getting post information for merited posts is not too difficult since the IDs are in the weekly merit.txt file (although you then need to scrape to get the post name, section/subsection, etc.), getting a complete history (or a date filtered one) for users is much more complex, especially if it needs to be done by scraping. Volume could be another issue too.

It can be done I figure, since those Bitcointalk phishing sites do a pretty good job of getting some of our posting history onto their site, but I can’t see myself doing it right now.

If would be a different story if posting information were to be provided by the Forum as files (i.e. post UserId, date/time created/modified, post id, post complete path or something of the sort).
Volume could become an issue even with this information (depending on what platforms one uses, since free/paid tend to have volume differences), but that is the least of the problems really.

To cut a long story short, I can’t really do that for now...

Given the incredible amount of daily posters we have, seems pretty low, but, also, given the incredible amount of daily shitposters, it seems symptomatic to me.

 It will be interesting to see an analysis related to the number of posts these users are making per week, or, if possible, in what boards are they being rewarded. Maybe it will give  a different point of view: for instance, if a user is creating more than a hundred post per week and receiving one merit, we can imagine the quality, or, on the contrary, if the user is like you, it will post about 20 per week and receive tons of merits just for one post. Besides, it should be interesting to see if the user is receiving merit per post (one per) or tons of merits per post (I figure it should be the stranger cases...). Maybe we can typify users by doing a personal analysis (by selecting... I don't know, 100 users?):
- Less merited ones- board they use- number of posts vs number of merits.
- The average of a merit per week -board more used- the number of post vs the number of merits.
- Very merited (more than 5 per week) - board more used - the number of post vs the number of merits.
- Pretty awesome (more than 10 per week) - board more used - the number of post vs the number of merits.
- Genious (more than 20 per week) - board more used - the number of post vs the number of merits.

Maybe just by making a selection of "types" or users we could have some answers.

Something that is complementary to my OP is getting a glimpse at how often forum members get merited. The idea that good posters will get an average of a merit per day or so seems more of an exception than a norm.
I decided to take a look at a more relaxed view, and see how many forum members are getting merited at least once per week. I’ve excluded from this view those forum members whose account were created after the Merit System kicked-in, simply to have a uniform set with the same possibilities (having an account older than 23 weeks).

The numbers that come out are as follows:
MWeeks represents the number of weeks that a user has been merited. Only 29 users (0,19%) of the merited user base have been merited every week (up until last Friday).
3,54% of merited users have been merited at least on 10 different weeks. But the gross 85,20% of merited users have only been merited on 1 to 3 different weeks.
As we can see, only very few are merited on a regular weekly basis. Now we cannot cross this information with the number of posts made per week, to make sure that users were actually posting each week, but it gives us a pretty good idea of how regularity seems to be working out.

Note: The above 15.011 merited users are those create before the Merit System was set in motion, having therefore had the chance to be merited each of the 23 weeks the system has been running. There are an additional merited 2.037 users that, as mentioned above, I’ve excluded from the above results for being created after Merit System kick-off.
If I were to include them in, the results are very similar and distribution variance is practically irrelevant.

They have actually made it via their helpful threads and topics, which are abundant in the forum.
The problem is, newbies who are mostly joining the forum to get free money and dont' care about knowledge, skills, will probably not want to learn from those sort of helpful threads and topics.
The only thing newbies (most of them) care about are bounties and money.

It's true. Honestly, I joined the forum with the original purpose like this.
Fortunately, after the launch day of merit system, I knew that I will never rank up if I don't change my minds, my writing approach.
I also realised that how bad I did before the launch day of the system and new rank requirements.
For now, I strongly believe that I have changed and become a more constructive member of the forum. 

With merit system, I think we all should give newbies time to realize what they should do, should change.
Over time, some of them will change as original purpose of merit system.
Of course, some of them will never change, but it doesn't matter of the forum.
We, constructive members, just simply ignore all shitty members, add them to Ignore Lists if we don't want to see their shitshows anymore. They will be stucked at low ranks.

Indeed, I agree with you. Higher ranks should guide the newbies and juniors. Why? Some newbies did joined bitcointalk forum because his/her friend earn lots of money then their mindset will put to earning money once they knew that when his/her rank increases so the money he/she will received then his/her mindset will change again to earning merits and that's happening already in this forum.

It's only natural that non sources will distribute their merit at lower amounts per post. However, sources definitely have a lot more leeway when distributing it. These two could actually complement each other quite well.

I'm pretty sure this is the case. However, I'm not sure what is the criteria for it. I imagine sources that are commonly running out of their monthly limit may have higher limits too.

We concur on the criteria principals, and the data just goes to show that this mindset is kind of necessary in terms of sMerit awarding. As I posted somewhere around here this week, the Merit Source monthly allowance is nearly the same as the total sMerit that transacted during last month, so something seems to be a bit off in the system right now. As you say, it would probably be good if Merit Sources were able to award most of their assigned quota (not by spending more time hunting for meritable posts, but meriting perhaps a bit more those that they merit with low values of sMerit).

P.D. I don't know if Merit Sources are all alike in terms of monthly allowance. I guess there are A/B/C quotas depending on multiple factors (i.e. frequented sections/subsections, time on forum, account age, etc.).

Jaaajajaja!!! I was about to merit you as well, for I do believe you are awesome, but, ok, I will save them for a deserving newbie!!!


Yes, we are used to talking about how newbies need to change their mindset in order to be a part of the forum, but we also need to change ours!! In my case, I have only few merits, so I will try to use them to help others to rank-up.
Absolutely agree with you.

I've recently had my head turned about distributing more merit per transaction. I've considered this several times before hand, but after listening to a few users that suggested that it would be better to reward higher amounts as opposed to smaller amounts spread among several posts would be more beneficial to the system. It seems that merit sources are typically running into the issue of not distributing all of the sMerit they received due to only rewarding smaller amounts across several posts, and then running out of posts to merit. Thus the merit is wasted. This months "pool" I've already eaten into, but that's just because I had several posts which were on the back burner that I intended on meriting. I think the next period will likely see me distributing higher amounts among posts which I've just started doing. If I run out of sMerit to distribute for a period I'll just put them in a list, and revisit them in another period when I've not used all the sMerit on current posts.

The reasons that I've considered doing this is because, of the amount of very high quality posts only receiving a handful of merit, because the general opinion is more is better. But, that's kind of shafting the users that have put high amounts of effort into a post to only receive 1 or 2 merits. When another post which is a quality post, but not to the standard of the previous one is getting around the same amount of merit. This is problematic, and higher quality posts should be standing out more among quality posts. I've got some rough guidelines which I'll now use to distribute merit. I originally had a maximum of 5 merit, but have now removed this as I think it's too limiting to those that have made exceptional posts. I've been starting to do it towards the end of my merit expenditure this period, but will certainly be rewarding higher amounts per post once the next period rolls in.

I've already made it clear that I don't really care about the rank of the person, and I won't be only meriting the lower ranks or saving it for lower ranks. A quality post is a quality post. Even, if they are Legendaries, Merit sources or even admins.


This is especially true when considering higher quality of posts. It seems that a lot of users are focused on finding posts which haven't been merited by others. This isn't a bad practice, and is certainly valuable. But, I think this damages the users which are spending a considerable amount of effort in their posts, and only receiving a few merits. It shouldn't really matter how many users have previously sent merit to the post before you consider sending merit yourself. If you think it's a quality post then you merit it regardless.  At least this will be the stance I'll be taking.

Dam it! I just merited your post since it shows the spirit and focus that I think we need to have. That’s two days is a row. The hand brake is off …

I will try too. I usually spend my merits as they come, but I will try to keep them to give an awesome newbie a great "punch". I do believe that by encouraging good posters we will be able of having really nice people around here.


Yes, but maybe we have settled the issue enough to have an argument in the case someone accused us or whatever. I think we have the same people in mind about the Spanish board, for this is easy (there are not too many good posters in there, but the ones deserving recognition are more than good, they're excellent!).


I guess the idea is that only good posters can rank-up, and that we are failing in that, to be honest, by giving merits 1 by 1. I think this is truly easy to detect the intentions of a newcomer by the way he/she is posting, for this is easy to see which ones are really interested on the forum and want to make it a better place by their contributions and which ones are just seeking for merits by making threads with "good appearance" but copying other's ideas.
From my perspective, the goals are clear: to encourage good ones to become an important part of the community and to relegate others to determined boards (meaning, of course, those "Forever Juniors" having fun in the Alts ones and coming in here from time to time just for cry about something), where they will be free of doing whatever they consider.
After reading the Theymos statement from the last week, I believe I understood the problem and his mindset. That's why we need to stop being shy regarding merit distribution and help the ones who deserve to rank-up. At the end of the day, to rank-up is a proof, as well as an example, of how far you can get by being commited to the forum's community. By encouraging some good newbies, by rewarding them, probably others will come.
To me, it is just the nicest way (personally speaking) of making from the good ones an example.

I believe that is what the data shows, and is potentially an issue. The vast majority of merited posts receive 1 sMerit and only get merited by one single person. That is more so in certain boards (like our local Spanish board) and a bit less in others.

I think an interesting question to reflect upon is this: would we still have the same awarding habits if suddenly we became “sMerit rich”? (not too rich, just wealth enough to avoid having to use each sMerit as if it were a gem).

In my particular case, I would alter my habits. Perhaps not on the amount of posts I merit (time is an issue and I’ve still got a few dozen sMerits to go - although at this rate I’ll finish them off soon), but certainly on the quantity I give per TX (I tend to do it in ones and twos so as to make them last, but would like to have a wider range of say one to five). I also refrain from meriting on occasions (not always) other merited posts, just because at least they got some and I don’t want to run out, saving it for later on for others.

Totally agree with you here.
 
I’ve just checked on the Merit Dashboard and was surprised to have awarded sMerit there to 19 different people. There is a select group of 5-10 that comes to mind and that really deserve frequent meriting due to the fact that their posts stand out with thought, but as you say, one tends to refrain from doing so since the circle of good posters is rather limited now days and it could seem like a closed-loop from an external observer’s point of view.

Here we would really need to get into the terrain of the “infamous” Forum Goals (“infamous” due to some posts and discussions on the matter that have taken place recently). Maybe the goals are set to have a very limited amount of users ranking-up per month. If that were true, the system is fine as it is since it meets those objectives.
Nevertheless, I doubt that the idea is to have members ranking-up at the current rate. Subjectively, I find the numbers really poor, and prone to be a problem in the mid-run. The idea is to get people to come to this forum and keep them around (with rules and a certain contributive spirit even if bounty hunting, whilst purging the spammers).

So numerically I cannot say really without knowing the goals (which we don´t and won’t, since it’s the “board of directors” role to set and monitor these). But my gut (inspired by the data) tells me that we’re low on all fronts. Perhaps I would vouch for more merits to be awarded to the good to best posters range (and I’m not talking just about the top 50 here), simply so as to facilitate their path if they deserve it. The forum cannot be happy with just the very best ranking-up (black sheep aside), or can it? …

Thanks for doing that. It illustrates really well our previous discussion on the Spanish board about how the merited posts are used to receive 1 merit by one person most of the times. Of course, there are exceptions, as you have presented, but those ones are few.
I've seen many times excellent posts with just one merit (if any), and, as we discussed the last week, I even stop myself of giving more than one merit (well actually now I don't have any, for I always give away each merit I earn), but even when I have some to spare I just prefer to give the merit 1 per 1 hoping that someone else give some other merit to the poster. In my case, I don't have many to spare, but, also, I have to confess, I rather give one than to being misunderstood or accused of giving merits to an alt I don't even have.
Regardless of the personal approach, I think the data is clearly speaking about the lack of merits. Sometimes I doubt for that to be the problem (especially when I search in the forum for meritable new users, really difficult to see), but I've come to the conclusion of the 50% reasons, meaning that one part of the problem is the lack of good posters and the other is just the lack of merits.
In the Spanish board, it is "easy" to earn a merit (one merit) by a person, if you post decently, but the problem is there are few in there and your merit history can be interpreted as farming, even when this is not. Besides, more likely you are going to obtain 1 merit from 1 person for an extraordinary post, this is difficult for the users in there to gain more. I think Paxmao is the only merit source around there and, of course, if one user is really good, he/she will obtain many merits from him, but the issue, again, is the risk of being misunderstood.

Well, that's just a personal approach.
Based on your data analysis (awesome as usual), Do you think the system needs more merit sources? Or just for the merit sources to don't be shy and give more amount of merits per person (if deserved)?

1.   Introduction
Recently I’ve seen various post refer to the number of sMerits been given out per transaction been low. This is an inherent feature in a system where the number of weekly awarded sMerits is pretty low on overall terms, and has stabilized as of late in the 3,9k-4,4k per week area.
I wondered how the number of sMerits per TX was behaving Forum wide, but was specifically curious as to whether this was roughly the same on all Forum Section/Subsections or not.

I also have the feeling that, regardless of the amount of sMerit been awarded per TX, the actual number of times a posts was been awarded by multiple users was rather low.This would lead to the two factors being combined in practice: low amount of sMerits per TX, and low amount of times a post is actually being merited…. And this is what is going on.
Now whether this is as it should be or not I cannot tell. But let’s speculate: if we were to magically double our sMerit over night or even triple it, how would we use it? Giving more sMerit per TX ? awarding more posts ? allowing ourselves to award sMerit to posts that we consider valuable, but that have already been awarded by some ? 
A noticeable increase in our sMerit accounts could probably change out awarding habits, but since that is just left there for speculation purposes, let’s analyse what reality has to offer…

Note: I’ve added the tabular data to the Merit Dashboard so that it can be filtered by different dimensions (dates, boards, etc.). Data is not graphical there.
This post aggregates all historical data. Anyone interested on narrowing down the date scope can do so on the Merit Dashboard (Post Summary tab).


2.   sMerit awarded per TX
I’ve simplified the cases shown here in order to avoid overcrowding the post with too many graphics. With that in mind, the information is seen from the overall forum’s perspective, and then is broken down by local/non-local boards for TXs that award 1 sMerit and TXs that award more than 5 sMerits (as I found these to be the most representative cases).

2.1 sMerit awarded per TX – Forum overall



Out of the (up to last Friday) 79.335 sMerit transactions (TXs), 69,08% award 1 sMerit, and another 14,29% award 2 sMerits. From there on the percentages decrease, with bumps on multiples of five.


2.2 Non-local boards – TXs involving 1 sMerit



If we take a look at the distribution of sMerit awarding per forum section/subsection, and focus on those TXs that are awarding 1 sMerit, we can see that there are big variations depending on the board that we focus on.
For example, Economics has 81,60% of the TXs being awarded to this board consisting of 1 sMerit. On the other extreme, Politics & Society has 54,23% of the TXs involving one sMerit.
This means that the awarding TXs are of a smaller quantity in Economics than in the Politics and Society. That is a 27,37 point difference between the two extreme board section.


2.3 Non-local boards – TXs involving >5 sMerit



Here too there’s a large gap between the board that has the greatest percentage of >5 sMerits TXs (Off-topic, being 11,19% of all TX on this board) and the almost lowest (Bitcoin Technical Support, with 1,09% of all the sMerit TXs in the board involving > 5 sMerits).


2.4 Local boards – TXs involving 1 sMerit



The Portuguese Local board has 88,15% of the sMerit TXs involving 1 sMerit, while on the Greek board this happens only on 50% of the sMerit TXs (but the absolute number is rather low there). The larger local forums (Russian and Indosenian) are in the 68-69% range.
In other words, the Portuguese board, along with Spanish, Japanese, German and Croatian, constitute the “meaner” boards in terms of sMerit awarded per TX. Of course the second derivative is that this is not some innate cultural feature, but rather more a lack of sMerit to award or a lack of good post to award to.


2.5 Local boards – TXs involving >5 sMerit



Here the distribution gap is also enormous. The Greek board has awarded >5 sMerits in 24,39% of the board’s TXs, while the Japanese only 0,4%. 60% of local boards are under the 5% mark.


 
3.   Number of times a post is merited

The same case simplification as above applies to keep the number of graphs to a reasonable representative level.

3.1 Number of times a post is merited – Forum overall



Out of the (up to last Friday) 55.893 merited distinct posts, 81,43% of them have only been awarded sMerit once, and an additional 11,40% twice.
We know that circulating sMerit seems rather low, but the fact that a community as large as this one has it’s meritable posts merited by one or two member in almost 93% of the cases is kind of low. It’s like saying the vast majority of the posts are only appreciated by 1 or two people ….


3.2 Non-local boards – Number of posts sMerited 1 time only



Bitcoin Technical Support comes out best, since 69,55% of the merited posts are only merited once (so the remaining 30,45% are merited more than once). The worst case is Politics & Society, where 91,05% of merited posts get awarded just once. That’s over a 20 point gap variation between the two extreme sections.


3.3 Non-local boards – Number of posts sMerited > 5 times



The absolute numbers are pretty low (see the lower row of the data table on the graph). Meta comes out best, with 3,65% of merited posts being awarded sMerit by over 5 people. On the other extreme, Speculation Altcoin has a 0,29% ratio, but even mining is really low (0,48%).


3.4 Local boards – Number of posts sMerited 1 time only



India has the worst ratio (Scandinavian is a real outlier), with 94,55% of merited posts on the board being merited only once. Arabic comes out best as ratio is down to 72,84% of the cases on its local board.


3.5 Local boards – Number of posts sMerited > 5 times



The Spanish board has a really poor ratio, since only 1 post (0,18%) has received sMerit from more than five people. Absolute numbers are so low, that I would only consider the Russian, Turkish, and Indonesian board ratios as representative, with less than 3% ratios but enough absolute base.