Steem Analytics - Is the distribution of Steem earnings changing over time?

Introduction

Six months ago I had a look at the distribution of earnings from the Steem blockchain. I analysed how much each account earned in author, curation and beneficiary rewards over a two week period and grouped accounts into buckets based on these earnings. The picture was fairly unequal, with a small percentage of accounts garnering a large proportion of the available rewards.

This week I revisited this analysis with the aim of understanding whether the overall distribution has changed over the intervening months and whether there is a trend of change in one direction or another.

For a recap, here is the original analysis:
https://steemit.com/utopian-io/@miniature-tiger/steem-analytics-distribution-of-earnings-1543522494404

0. Data

In the new study I have created a distribution of earnings for each month from September 2018 through to April 2019 and looked at the pattern of change. Each monthly analysis uses the earnings from the first two weeks of the month, since I believe this is sufficient data for stable results. I initially looked at the STU earnings, as per the original study. I then changed to Steem earnings, since this provides a more valuable comparison over time without the impact of the change in the Steem price.

I have excluded the results from October 2018 from consideration (although I have included them in the charts). After the HF20 on September 25 2018 there was a period of 5 days without voting and then a couple of weeks where the subsequently overflowing reward pool was emptied. This led to a much greater distribution of Steem over the first two weeks of October 2018 and distorted the distribution for that month.

1. STU earnings: September 2018 vs April 2019

What did the STU earnings distribution look like back in September 2018?

In summary we have:

75% of users earning less than $1 STU in total over the two-week period and taking home 1% of total rewards.
17% of users earning up to $1 STU per day, i.e. up to $14 STU for the two-week period and taking home 9% of total rewards.
6% of users earning between $1 STU and $5 STU per day and taking home 22% of total rewards.
2% of users earning more than $5 STU per day and taking home 68% of total rewards.

How has the STU earnings distribution changed in the seven months to April 2019?

75% of users earning less than $1 STU in total over the two-week period and taking home 1% of total rewards.
17% of users earning up to $1 STU per day, i.e. up to $14 STU for the two-week period and taking home 10% of total rewards.
6% of users earning between $1 STU and $5 STU per day and taking home 24% of total rewards.
2% of users earning more than $5 STU per day and taking home 65% of total rewards.

The answer to the question posed is "not very much". There appears to have been little change.

However one significant distortion when using STU for the comparison over time is that the price of Steem varies considerably. For example it was around 70-80c in September 2018 and 40-50c in April 2019. As such our earnings buckets for the two periods are not really comparable. We would expect more users in the top bucket when the price of Steem rises and fewer when it falls. But this does not really tell us if the underlying distribution of earnings is changing.

Since the amount of Steem distributed is (broadly) level over time it would be more interesting to look at the change in the distribution of Steem.

2. Steem earnings distribution: User count

How has the distribution of earnings in Steem changed over the last eight months?

Let's start by looking at the percentage (and number) of users in each earnings bucket over time.

Steem has been valued at less than 1 STU across the analysis period. As such I have separated the highest earnings bucket into two. We now have the following earnings buckets:

Less than 1 Steem in total over the two-week period.
Up to 1 Steem per day, i.e. up to 14 Steem for the two-week period.
Between 1 Steem and 5 Steem per day, i.e. a maximum of 70 Steem for the two-week period.
Between 5 Steem and 20 Steem per day, i.e. a maximum of 280 Steem for the two-week period.
More than 20 Steem per day.

The first chart looks at the count of users in each earnings bucket over the last eight months (similar to the chart with blue bars in section 1 but with Steem earnings rather than STU earnings).

I think the main conclusion is that the results are really very stable across the eight months. In every month around 70% (69% - 74%) of users are in the lowest earnings bucket with less than 1 Steem in total earned over the two week period. The "1 Steem per day" bucket varies in a narrow range between 17% and 19% and the third bucket between 6% and 8%. It does not appear that much is changing.

There does seem to be a very slight trend of user numbers passing from the lowest earnings bucket into higher earnings buckets. However, this can be investigated further by looking at the actual user numbers rather than the percentages, as shown by the chart and table below.

This chart suggests to me that the trend is most likely due to lower numbers of newer/smaller users in 2019 compared to 2018, following reductions in the Steem price.

3. Steem earnings distribution: Value of Steem in each bucket

The chart looks at how Steem is distributed across the different earnings levels by amount. The chart uses the same buckets as the chart above but aggregates the amount of Steem in each bucket rather than the count of users.

Again, the percentages are pretty stable. There does appear to be a slow trend to the right, with a reduction in earnings for the lower earnings buckets and an increase in the earnings for the higher earnings buckets.

This could be evidence of higher earning users gaining a larger slice of the pie. However the results by value rather than by percentage show that there is also an increase in the user count for the top bucket.

The average earnings per bucket show lower increases for the top bucket.

It is worth bearing in mind that over the last year we have seen a rapid rise in dApps and vote bots on Steem and that these represent many of the highest earning accounts. One drawback of the analysis is that it does not consider the full picture once the earnings of these dApps and vote bots have been transferred back to the (most likely, smaller) users that provided the delegated Steem Power.

As such what we may be seeing here in the top bucket is an increase in the number of successful Steem businesses. The full picture of the distribution may be different once the earnings of these businesses are transferred back to the delegators. Further study would be required to illustrate this.

4. Conclusion

Overall I would conclude that:

The distribution of users at each earnings level is really very stable across the eight months. Things do not appear to be changing significantly.
The distribution remains unequal, with 70% of users receiving less than 1% of rewards, and 1% of users receiving 50% of rewards.
For the distribution of value, there does appear to be a slow trend of increased earnings for the highest earnings category. I believe that this is due to an increasing concentration of Steem Power in Steem businesses such as dApps and vote bots.
The full picture of the distribution may be different once the earnings of businesses are transferred back to the delegators. Further study would be required to illustrate this.

5. Notes

This study is a particularly tricky one and the results should be considered as broadbrush / indicative rather than absolute. Examples of difficulties include:

Vote bots: I have reduced author earnings by the value of the votes purchased from vote bots, rather than by the cost of the votes. Any profit / loss gained by users is thus excluded from the analyiss.
I have captured all main vote bots using the account names but I have not captured vote purchase systems where the votes are made by secondary accounts (e.g. minnowbooster). This will distort the authors earnings.
Some situations are just complex. Users purchasing votes and setting a high beneficiary percentage to an alt account will end up with a loss on one account and a profit on another. Similarly heavily flagged users with vote purchases may end up with losses. I have excluded any accounts making losses over each two week period. The number of these is small and I do not believe these distort the picture unduly.

Repository:

https://github.com/steemit/steem

This analysis is of data from the Steem blockchain which is an open source project.

Tools and scripts:

I used the block.ops analysis system to produce this study. Block.ops is an open-source analysis tool designed for heavy-duty analyses of the Steem blockchain data.

You can find the repository for block.ops here:
https://github.com/miniature-tiger/block.ops

The analysis used all the Steem blocks from the period analysed.

The study can be recreated by (once I upload the new analyses to github!):

Loading the data for the relevant time period into block.ops.
Using the earningsdistribution command from the command line, for example:
$ node blockOps earningsdistribution "2019-04-01" "2019-04-15"

Thanks for reading!

Thanks @crokkon!

What you mention in a side note here is a pretty complex task on its own but also highlights a central aspect of Steem when it comes to author reward analyses. I agree with your choice of excluding the bots here, even though it's almost impossible to exclude all paid votes.

Yes, untangling the web of Steem transfers, delegations, payments and vote purchases is very, very difficult, so the results should be considered as indicative rather than absolute, particularly at the top end. The lowest bucket results should be pretty concrete.

It may be possible to use the net transfers to/from vote bots and dApps to capture these elements but this is not without difficulty. Maybe I will study these elements independently first to see if I can add them in.

Dapps have another good share - I wonder to which buckets their votes mainly go into?

I counted 15 "Utopian usernames" in the top bucket for March, so with 315 Utopian contributors supported in total and a fairly long tail on the Utopian vote distribution I would guess Utopian users are spread across the top four buckets, and mainly in buckets 2/3. Smaller SP dApp users are probably concentrated more in the middle three buckets.

Thanks for the review!

Sort:

Trending

[-]

crokkon (69) 5 years ago (edited)

$5.96

18 votes

miniature-tiger (70) 5 years ago

$0.15

2 votes

utopian-io (71) 5 years ago

Thank you for your review, @crokkon! Keep up the good work!

$0.00

improv (71) 5 years ago

You do such cool work. I had no idea when we were teammates in steem-pocalypse that you were a numbers wonk! It's wonderful!

$0.03

1 vote

There's so much interesting data to look at!

steem-ua (64) 5 years ago

Hi @miniature-tiger!

Your post was upvoted by @steem-ua, new Steem dApp, using UserAuthority for algorithmic post curation!
Your post is eligible for our upvote, thanks to our collaboration with @utopian-io!
Feel free to join our @steem-ua Discord server

Hey, @miniature-tiger!

Thanks for contributing on Utopian.
We’re already looking forward to your next contribution!

Get higher incentives and support Utopian.io!
Simply set @utopian.pay as a 5% (or higher) payout beneficiary on your contribution post (via SteemPlus or Steeditor).

Want to chat? Join us on Discord https://discord.gg/h52nFrV.

Vote for Utopian Witness!

steemitboard (66) 5 years ago

Congratulations @miniature-tiger! You have completed the following achievement on the Steem blockchain and have been rewarded with new badge(s) :

You received more than 5000 as payout for your posts. Your next target is to reach a total payout of 6000

_{You can view your badges on your Steem Board and compare to others on the Steem Ranking}
_{If you no longer want to receive notifications, reply to this comment with the word STOP}

Do not miss the last post from @steemitboard:

New japanese speaking community Steem Meetup badge