Bitcoin Forum
November 04, 2024, 10:44:34 AM *
News: Latest Bitcoin Core release: 28.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: Bitcointalk – Posts per Day - evolution during past six months  (Read 578 times)
DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
July 23, 2018, 06:42:43 PM
Merited by suchmoon (7), qwk (5), dbshck (5), theyoungmillionaire (2), ABCbits (1), bitmover (1), Don Pedro Dinero (1), dwwbit (1), Helana (1)
 #1

1. Introduction:

The Merit System has been going now for six months, and I wondered how that correlated to posting and its evolution during those six months. The official forum stats (https://bitcointalk.org/index.php?action=stats) indicate that the average posts per day is 8.272, which always struck me as a smallish figure for all the posts we see day to day, and the amount of those that we visually classify as spam. The figure itself seems manageable easily from the merit awarding point of view, even more so considering that spam is likely to be the majority and can therefore be dismissed in a second whilst revising posts to merit.

It turns out (at least from what I can gather), that the number given as average posts per day is really quite off from what other data suggests. Instead of being 8.272 post per day, the last six month give me somewhere around 58.733 posts per day, with around 41.782 post per day during the last month. That is way more than the average states. I figure that the average shown on the official stats shows an average over a longer period of time (perhaps all history of Bitcointalk). Of course I may be wrong, but I’m basing the above on other data available through stats on the forum (using the post counts shown on https://bitcointalk.org/index.php for a specific date, and comparing that to the same data a month ago, extracted from WayBackMachine https://archive.org/web/).

The data per subsection I believe adds up the Child Boards too.


2. Post Distribution per subsection

The overall historical accumulative distribution of posts per subsection is as follows (derived from https://bitcointalk.org/index.php):



We can see that Ann Altcoins accumulates 19,68% of the posts, followed by Marketplace 17,69% and Marketplace (Altcoins) with 12,76% of all historical posts.
One thing to bear in mind is that deleted posts are not considered here (there is no stat on deleted posts), so they are not cannot be accounted for on this analysis. They do create some strange situations, as we’ll see in the following sections.


3. Number of posts and distribution per subsection – 6 month evolution

Using the information derived from https://bitcointalk.org/index.php for yesterday and retrieving the same statistic from the WayBackMachine at approximately one month intervals (there is no snapshot of the statistic for every single day, so I chose the best suited to aprox. 1 month intervals), we get the following table of data:




The above table is used to derive all the remaining charts and information. The T.Posts shows us the Total number of Post accumulated until that day. I’m not going to focus on this table as it is, but a derived one which is easier to follow. I’ll leave the above as a reference to the data origins.



The above table is much easier to follow. It shows the number of average posts per day per subsection, which is a good indicator to the activity going on in the subsection.

The interpretation is as follows:
Month 1: period of time between 23/01/2018 .. 23/02/2018. The number shows the number of posts per day for that period (calculated substracting the number of posts on the two dates of the interval and dividing by the number of days in the interval).

For example, Bitcoin Discussion Month 1 has an average of 1.806 posts per day. That is calculated by substracting the read on the 23/01/2018 (1.614.537) to that on the 23/02/2018 (1.670.511), and dividing by 31 days between the two dates.
And so on ..

I’ve marked some of the number in yellow, since they are inconsistent for being negative. What it really means I guess is that, for the given subsection and month, there must have been a purge of posts leading to mass deletion or movement to archival. At least that is my guess. This can also affect drops in the numbers, and we cannot know exactly to what degree, but it is a rather good approximation to what is going on.
Nevertheless, since the average posts per day is provided for 6 reads for each subsection, I think we can get a pretty good idea of how it is evolving:


3.1 Bitcoin – Posts per day


Mining is very odd, and it looks like a lot of posts have been deleted or moved to another section. All other sections seem to have a descending posts per day trend.


3.2 Economy – Posts per day


Here again Marketplace is very odd, and looks like it has been purged over the last couple of months. Economics and Trading Discussion are on the rise.


3.3 Other – Posts per day



All sections are on a downward trend. Meta has dropped by half what it was six months ago (the first month of the Merit System I guess). Off-topic’s downfall is very abrupt between the first month and the second, and second to third. Politics & Society takes a dep fall too. Looks like a clean-up too.


3.4 Alternate Cryptocurrencies – Posts per day


Post per day are massive on Marketplace (Altcoins), Altcoin Discussion and Ann, with Speculation (Altcoins) being the only one with an upward trend.


3.5 Local Boards – Posts per day




Most local boards also have a downward trend in terms of posts per day. This trend is rather steep in the Indonesian local board, and the fall on the Chinese seems too large to be real. These are probably affected by many posts/threads being deleted.

All these downwards trends could be due to:
-   More activity on the reporting and deleting front of posts/threads.
-   Less user activity per-se.
-   Less active users (?).
-   Summertime effect (recent month).

All in all, it looks like there are less posts created per day, since the overall average has gone down from 63K to 42K posts per day last month. This really depends on how much deleting has been taken place though.
That is still miles away from the average posts per day on the official stat which is 8,3K posts per day.


4. Merit per Post Ratio

With the previous data, we can derive the average merit per post for all forum Subsections. This may be useful to compare how easy/difficult it is to get merited in one section vs another.
The ratio is constructed simply by dividing the total number of posts by the total sMerits awarded to posts in the section for a given period of time (6 month; last month).
Now obviously this is not the probability of getting merit on a given post or section (as it mixed all sorts of posts with no filter), but from a macro perspective, it could allow us to compare sections based on this indicator.

For example, the Spanish Local ratio is now of 0,033, while the Italian is 0,065 and the Portuguese 0,036. This means that both the Spanish board and the Portuguese board have the same ratio of merit per post, while the Italian is nearly double. Looking further at the data on the table, the Spanish board made nearly twice as many posts as the Italian last month, and got nearly the same amount of merit (thus the ratio is half of the Italian board). Italian Posting and Portuguese are alike though in volume, but ratio is quite different between them.

This has to be taken with care, since it does not account for quality of the posts, how many are spam, etc. It is at least a curious exercise to do.




bitart
Hero Member
*****
Offline Offline

Activity: 1442
Merit: 629


Vires in Numeris


View Profile
July 23, 2018, 08:35:21 PM
 #2

I hope this means less and less spam as the months goes by...
Some data would be useful about the deleted posts (maybe someone from staff could create a similar table) that piece of information would let you to analize the causes of the post count decrese even better.
Some data would also be useful about the report statistics, to check if it's in line with the deleted post count.
Thanks for the stats and the tables!
DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
July 24, 2018, 04:02:16 PM
 #3

<…>
Some data would be useful about the deleted posts (maybe someone from staff could create a similar table) that piece of information would let you to analize the causes of the post count decrese even better.
Some data would also be useful about the report statistics, to check if it's in line with the deleted post count.
<...>
Yes, official stats on deleted posts and banned/nuked accounts per month would be nice to have around and you’ll help not only to counterbalance and interpret the drop in posts per day shown on the graphs, but also to measure and state clearly the results of all the reporting being done by forum members.

@Cabalism13 opened a post today commenting on how the forum seems quieter now, which is rather much in line with what the above stats show, although the average posts per day is still pretty high and above the average official count.

TomCrypto
Full Member
***
Offline Offline

Activity: 715
Merit: 220


View Profile
August 10, 2018, 12:12:50 PM
Merited by DdmrDdmr (1)
 #4

Thank you OP for this really interesting post as always with you Smiley

Concerning the reason behind all these downward trends I would think also of another possible reason (related to reasons 2 and 3 you mentioned) for this : The bear market.
Indeed since the beginning of the year all crypto are in a downtrend. Due to this downtrend, people are less interested by the cryptoworld than before. As you can see on Google Trends for the word bitcoin in the last 6 months you obtain the following graph :

Due to this lack of interest, there is less new users that come to crypto due to less Hype, older users already into the cryptoworld tends to trade less/invest less/check less there portfolio or just wait in cash. A lot of people are starting to loose interest for the time being on the crypto world and they follow it less closely and just wait for the storm to pass. 
Therefore the same apply for Bitcointalk forum in my opinion. With the bear market less forum users are posting because they just wait, there is less projects with huge hype because it’s not a good time to invest (not everyone follow Baron Rothschild advice about investing “when there is blood in the street”), the project bounties are paying less because at the end of the projects ICO their token are not worth a lot due to bear market. All of this will surely generate less post in BTCT forum in my opinion. When there will be confirmation that the bear market end, 2 months later I’m sure that the number of posts will increase a lot! But I could be wrong Smiley
DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
August 10, 2018, 12:54:58 PM
 #5

<...>
Yes, I figure you're there. It’s a combination of factors that add up and justify the predominant downward trend (less hype, reporting/cleaning, merit barrier, etc.). It would be interesting to do this exercise after a few months of sustained bullish market. We can get back to this when this happens.

By the way, nice to see that posts can still be found by others after some time has gone by (I would have guessed that the OP was probably on page 10-15 by now of Meta).
TomCrypto
Full Member
***
Offline Offline

Activity: 715
Merit: 220


View Profile
August 10, 2018, 02:14:09 PM
 #6

I can't wait to see the result of the same work in a bullish market!

It was quite easy to find this post because i like to follow the work you do on data analysis and therefore I watch the posts you start.
I even do some advertisement of your work on the french local board :p (https://bitcointalk.org/index.php?topic=2820262.msg37494332#msg37494332 / https://bitcointalk.org/index.php?topic=2820262.msg43825960#msg43825960)

DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
August 10, 2018, 02:21:55 PM
 #7

<..>
It was quite easy to find this post because i like to follow the work you do on data analysis and therefore I watch the post you start.
I even do some advertisement of your work on the french local board <...>
Ah, and just when I had high hopes that posts maybe don’t get too buried in the end really …
I often find that there are many statistics, posts in general, that could be interesting around, so giving them visibility through local translations is good for the non-English speakers to gain access to information. Cheers.
Helana
Member
**
Offline Offline

Activity: 80
Merit: 71


View Profile
August 23, 2018, 04:46:50 PM
 #8


4. Merit per Post Ratio




This is such an awesome work... yes, you actually helped me a lot to understand the statistics and how merits are distributed. Taking a look at your last graphic, it seems like the ANN board is well-merited. I was just interested in that because I've seen a lot of people complaining about the merit system, and I couldn't understand why.
Well, the numbers are speaking: under my interpretation, actually, the ANN board is the most merited, but also one of the harder places to obtain any merit. So, please, correct me if I'm wrong: the most merited board - in relation with the number of posts, of course- is Meta, followed by Development and Technical Discussion and Bitcoin technical support? I'm not a number genius, but it seems that the merits per board actually are speaking about the system as a pretty fair one. I mean: the boards that actually are concerned about the forum and in which the most interesting threads, the most helpful ones and also the most informative one -like this one- are then the most merited per post.

So I guess that everytime the people write: "you need to create an informative and helpful post in order to get some merit" they can actually demonstrate it by using this very stats.  Somehow I did read this statement so many times, that it began to sound quite like a cliché and I was also asking myself if there is any way to demonstrate it. But, actually, there is one.  Smiley

On the other hand, this is kind of disturbing how many deleted posts are being merited -this information belongs to your other amazing post: https://bitcointalk.org/index.php?topic=4932684.0  - and also how the ANN section can be actually using merit as a way of payment? Of course, I'm sure there are a lot of informative threads deserving merits on this one, but, I couldn' t find many so far, so, can we speak about two different ways of using merits?:
1.- Like actually a recognition.
2.- Like an asset: people giving merits to others in order to receive something in return?

DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
August 23, 2018, 06:20:03 PM
 #9

<...>
Well, the numbers are speaking: under my interpretation, actually, the ANN board is the most merited, but also one of the harder places to obtain any merit. So, please, correct me if I'm wrong: the most merited board - in relation with the number of posts, of course- is Meta, followed by Development and Technical Discussion and Bitcoin technical support?
<…>
Correct, but you skipped Serious Discussion which is actually above Meta in terms of merit per post.
As you’ve seen on the Ann thread for example, the most merited overall subsection may not bare one of the best merit to post ratios at all; quite the opposite. What the numbers cannot say is precisely how many of the posts in each section are reasonable candidates to merit and which are not (until some brilliant AI comes along that analyses post content).
Logically (and visually), subsections such as Meta, Serious Discussion and Bitcoin Technical Support have better content and way less spam than Ann subsections, therefore the candidate posts to merit are a larger proportion, but this factor as I said, being a very relevant one, cannot be measured but only perceived.

Quote
<…> On the other hand, this is kind of disturbing how many deleted posts are being merited <…>
Actually, the order of factors is the other way round: a post is merited first, and then deleted (we cannot merit a deleted posts, because it is not there anymore). Yes, many merited posts do get deleted for all sorts of reasons: Ann threads (or others) that are trashed, merit abuse cover-ups, reporting posts that are awarded merit but later deleted as a way to cleanse the list of reporting posts dealt with (this seems legit, and was a practice that due to the attention it drew has now been abandoned I think), etc. Any forum rule that is broken by a post/thread can get the post/thread deleted, and so we lose track of which post the merit went to – but at least we can quantify it and track it.

Quote
can we speak about two different ways of using merits?:
1.- Like actually a recognition.
2.- Like an asset: people giving merits to others in order to receive something in return?
It should be in the former area for the most, especially now days that there is less merit in circulation. Eventually, people are allowed to merit whomever they wish and for whatever they want, although there is a spiritual driver for the system: quality/interest. When awarded merit differs from this, is when we get suspicions of merit abuse going on behind the scenes …
l3pox
Legendary
*
Offline Offline

Activity: 2702
Merit: 1170


There's no need to be upset


View Profile
September 16, 2018, 08:38:40 PM
 #10

thanks a lot for putting this data together @DdmrDdmr
the last table is specially valuable to understand better the merit system
how did you pull all the data? with a programming script?

as stated by @Helana the last table is specially valuable.

this thread was mentioned as well on the Merit & new rank requirements and again by me

as stated:

most merits per point are given on the following subjects (ordered):

Serious Discussion
Development and technical discussion
Meta
Bitcoin Technical support

(on the month of june, but with simmilar data to the last 6 months before it (first semester of 2018)

███████████████████████████
███████▄████████████▄██████
████████▄████████▄████████
███▀█████▀▄███▄▀█████▀███
█████▀█▀▄██▀▀▀██▄▀█▀█████
███████▄███████████▄███████
███████████████████████████
███████▀███████████▀███████
████▄██▄▀██▄▄▄██▀▄██▄████
████▄████▄▀███▀▄████▄████
██▄███▀▀█▀██████▀█▀███▄███
██▀█▀████████████████▀█▀███
███████████████████████████
 
 Duelbits 
██
██
██
██
██
██
██
██

██

██

██

██

██
TRY OUR UNIQUE GAMES!
    ◥ DICE  ◥ MINES  ◥ PLINKO  ◥ DUEL POKER  ◥ DICE DUELS   
█▀▀











█▄▄
 
███
▀▀▀
███
▀▀▀
███
▀▀▀
███
▀▀▀

███
▀▀▀
███
▀▀▀
 
███
▀▀▀

███
▀▀▀
███
▀▀▀
███
▀▀▀
███
▀▀▀
███
▀▀▀
 
███
▀▀▀
███
▀▀▀
███
▀▀▀
███
▀▀▀

███
▀▀▀
███
▀▀▀
 
███
▀▀▀
███
▀▀▀
███
▀▀▀

███
▀▀▀
███
▀▀▀
███
▀▀▀
 
███
▀▀▀
███
▀▀▀

███
▀▀▀
███
▀▀▀
███
▀▀▀

███
▀▀▀
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀
 KENONEW 
 
▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄
▀▀█











▄▄█
10,000x
 
MULTIPLIER
██
██
██
██
██
██
██
██

██

██

██

██

██
 
NEARLY
UP TO
50%
REWARDS
██
██
██
██
██
██
██
██

██

██

██

██

██
[/tabl
DdmrDdmr (OP)
Legendary
*
Offline Offline

Activity: 2492
Merit: 11048


There are lies, damned lies and statistics. MTwain


View Profile WWW
September 16, 2018, 09:03:27 PM
 #11

<...>
the last table is specially valuable to understand better the merit system
how did you pull all the data? with a programming script?
<...>
It’s a combination of data origins: the posts per day per subsection are derived from the post count that exists on the forum itself for each subsection. To get retrospective data to compare monthly increments, I used the WayBackMachine to retrieve the same data from an archived snapshot. Copy those figures to excel or so and you’ve got part of the data. Getting the merit awarded per subsection in a given timeframe is more complicated: you need to build (properly) your cumulative database from all the merit.txt official weekly files, and then scrape and cleanse at least the section/subsection for each merited post from the website itself. Then it’s down to turning data into information that may seem meaningful...
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!