STEEMIT Usage Stats for the Last Month, and How To Improve Them

 I have been monitoring the data from the service that shows user  activity and their share in STEEMIT, and I would like to share my  observations with you.

The first thing that I was interested  in is the user base growth and how active it is. I herd that STEEMIT is  growing rapidly and even surpassed early Facebook in that area. Well,  here are the real graph for the last month.

As  we can see from the graph, the amount of users has grown significantly,  but the weekly and hourly activity hasn't actually risen, and even fell a  bit.

What could that indicate?

1) It could indicate that measures against bots is bearing fruit, and bot activity 

2)  Since we do not know the ratio between bots and real people, I suppose  that the lack of growth of active users is connected with the fact that  after comming to STEEMIT they do not find the reasons to stay.

Let  us try to understand why that is happening, and what should we do about  that. 

The experience of some of my acquaintances is that people usually  come to STEEMIT in reaction to the information that this is a social  network that offers an opportunity to earn something by actively  participating in it. But they quickly become disillusioned, because in  reality for about 90% of the users it isn't possible.

I  graphed the Lorenz curve for the STEEMIT users. It is usually based on  the income of the population, but I based it on the share of the STEEMIT  assets. Here is the result. 

The complete and  total inequality between the STEEMIT users is absolutely obvious, there  is no such imbalance even in the most economically backward countries of  the world. What does that leads to? 

1) The probability to  earn something for their content is minuscule for 90% of the users,  based on the current reward algorithms. Due to this huge imbalance only  few authors that could attract the whale votes can earn something, and  thus a huge amount of quality posts sink to the bottom without earning  anything. There are a lot of examples of quality content posts earning  around ten cents, and posts with some rubbish rise hundreds of dollars. 

2)  New users can't earn by curating. Current curation reward algorithm  doesn't allow low SP users to earn the reward, due to the fact that user  weight matters, but the place in the voting queue almost doesn't  matter. Lets use me as an example. If I, currently with 200SP, vote  right before a whale in a post that pays out between 400-3000 dollars,  my reward would be between 0,001 and 0,007 SP.

The problem is  that voting after a whale is meaningless, but the popular posters have  bots following them that vote between 5 and 30 minutes and dilutes the  votes of the small SP users that vote before the whales. The problem is  that searching for the quality content that is going to rise is almost  impossible, since it involves deducing the decision making processes of  the whales, but people would rather make their own decisions. 

It would seem that dilution of the whale share would solve this problem by giving more weight to the common users, right?

So, how does this dilution process is happening over time? Here is the monthly data for all user base.

The categories of users that I used are the same as in https://steemd.com/distribution ( FYI: the @steemit account is excluded from the statistics ) but the newbie and user levels were grouped as "plankton". If I were to  track newbies separately, the results would be even sadder. It should  be noted that the change between user categories is happening, but the  status quo persists.

So, how did the shares  change? The whales lost 2,91%, the orcas gained 1,46%, dolphin share  gained 0,75%, minnow share increased by 0,24% and finally the plankton  gained 0,27%. It would appear that things are progressing swimmingly  (pardon the pun) with whales losing almost 3%, and the monthly numbers  are good. But the share of the biggest segment had the smallest  increase, which means that for 90% of the people nothing has changed at  all.

Lets check the numbers for the users that were active in the last week.

Things  are even more peculiar in this dataset. In the following table I assume  the assets of the active users as 100%. That results in whale share  shrinking by whooping 5,62%, orca share increased by 3,41%, dolphin  share increased by 1,78%, minnow share increased by 0,44%, but the  plankton share dropped by 0,02%.

In my opinion  we have a problem that needs solving, since the minnows are the main  body of the new users of STEEMIT, and if the capital doesn't flow into  that segment, then STEEMIT will turn from the social network of the  whales into the social network of the whales and a couple hundred of the  author that those whales are subscribed to. Of course I am  exaggerating, but without more active flow of the assets into minnow and  plankton segment STEEMIT will not become a normal social network, where  the content is selected by all of the user base, and will remain the  social media for the chosen elites.

What should be done to drive the assets towards the plankton?

1)  The whales should continue Powering Down. It depreciates the value of  STEEM tokens, but allows the whale share to decrease relatively quickly.  And allows people to buy the inexpensive STEEM to increase their  weight. But this process will never be large scale for the new users, so  we need to consider another mechanisms.

2) The  curation rewards should be changed. In my opinion right now it is only  possible for the whales and bigger orcas to get earn by curation, and it  is apparent that even than this process is quite specific: when we have  a whole group of authors that are on the lists of some whales, so they  consistently rise to the top, and the orcas are also trying to  capitalize on that, by upvoting before the arrival of the whales. There  is no place there for the small weighted users there. 

What needs to be changed in the curation reward algorithm (those are some variants for the discussion):

A)  decrease the curation reward of the popular authors. Right now the  whales are subscribed to those authors, that upvote their content  automatically. There are also a lot of people who are trying to earn by  upvoting before the whales, to earn something by curation. The current  system is not helping the goal of promoting the new authors with quality  content, because everyone is trying to earn by curating the already  well known authors.

If the curation reward for  curating well known authors would start to decrease, then the curators  would have to constantly look for quality content in other places. The  existing popular authors would lose nothing, because the users would be  thanking them for the quality content, not for the curation rewards.

B)  decrease the dependency between the share of the curation reward on the  asset weight of the user. Right now the small users are hampered by the  30 minute 'bot limit' and the fact that by 20th minute their share in  the upvote would be almost equal to zero. I suggest at least temporarily  decrease the dependency of the user weight in the share of curation  reward and increase the significance of the position in the upvote  sequence. This would allow small users to start earning at least  thousandth of SP, which would increase the weight of the plankton in  total.

C) My opinion is that that the author reward should be temporarily changed. 

As it is known the post reward is distributed approximately like this:

The  reward is proportional to the ratio of votes[x]2 / sum(votes[0...n]2),  since the whale votes are very large compared to the other users the  quadratic dependency only increases the imbalance. Yes, this is a good  idea that works when there is a greater equality between the vote power  in the system, but right now it prevents the content that was overlooked  by the whales, but was noticed by other users from earning something.  Thus a temporary abandonment of such dependency in favor of proportional  would allow the good authors that the whales aren't subscribed to earn  more.

I'd like to hear your opinion in the comments.

This is a translation of this post by @xanoxt, 10% of SBD reward will go to him for the service.