Witnesses play a vital support role in keeping Steemit up and running. And each witness relies on votes from steemains to keep their role as a witness active, just like us steemains need votes on our posts to keep us as active authors.
With this in mind, I decided to analyse the witness accounts and the voting habits of the top 50 witnesses.
Approach
Data Transformations
Using Power BI I connected to https://steemit.com/~witnesses and pulled down a list of the top 50 witnesses.
I then also used Power BI to connect to steemsql held and managed by @arcange . The first table I connected was the TXvotes table. I used the listed pulled down of the top 50 witnesses to filter the TV votes table and I also filtered the table to look at votes made in September only.
Next I connected to the accounts table. using the voter name from the TXvotes table, and the account name on the accounts table as common fields in both tables, using an outer join I merged the two tables so that the account creation date, account reputation and vesting shares are now showing the TXvotes table to give greater further details on the witnesses.
Finally I connected to the Comments table in steemsql. By using the permalink field in the comments table and the permalink field in the TXvotes table as the common fields, I carried out another outer join to pull in the category for the post and the post depth.
The final steps in the data transformation were to change the date fields from data type date and time to data type date. This is an important step for time intelligence functions to work when modelling the data.
The only tables loaded to the data model is the Witenss list table and the newly expanded TXvotes table.
Data Modelling
I use DAX (Data Analysis eXpressions) to carry out calculations across the tables of data. The first step in modelling the data was to add a date table. In Power BI a date table is required for time intelligence functions to work. After this I carried out further calculations on the data which are represented in the visualisations below.
Who are the top 50 Witnesses
On average, witnesses in the top 50 have been on Steemit for over 440 days. There are some exceptions to this. In number 50 is @netuoso, who is only on steemit 96 days as at the date of analysis. On the opposite end of the scale we have @blocktrades in number 9 and that account was registered 540 days ago.
The average vesting shares held in the by the top 50 wintesses is 492.87M, but again there is a huge variation between those at the top of the scale and those at the bottom. @Blocktrades is well ahead with 7469M vests whereas @chainsquad.com has only .814M. This is not an indication of how much they have made here on steemit, as for this one would also need to look at withdrawals. I have not included withdrawals in this analysis.
Finally I had a quick look at posts to see how active witness are when it comes to posting. I was very surprised to see two witnesses with no posts at all. It makes me wonder how they are in the top witnesses? If they are not interacting with the community who is giving them votes and why? Hmm that’s another analysis. Besides that, on average witness have made 1660 posts each.
The visualisation below plots the post count by the number of days on steemit for each witness and the size of the bubble represents the vest shares held
Voting
In total the top 50 witnesses voted 177,493 times. Of this only 0.2% of votes were self-votes and 0.9% were votes to other witnesses.
First I sorted the witnesses by the number of votes given to see who gave the most votes
And then by average weight to see which witnesses vote with a lot of power
Next I had a look to see which witness votes for themselves the most
And who votes for themselves with a lot of power
Sorting by the number of votes given to other witnesses in the top 50
The visualisation below plots the average % weight for self-votes, against the average % weight of a vote to witnesses and the size of the bubble represents the total average weight of a vote given by the witness.
When are witnesses voting?
The 12th Sept has been the busiest day that witnesses have voted so far, with the average daily vote reducing in the last 6 days. The busiest time seems to be at 9:25 am however the busy period is from 7:00am till 17:00pm
What do witnesses vote for?
Analysing the categories on posts voted for by whales, I first looked at the number of votes given to each category
Then I looked at the number of posts voted on in each category by witnesses
I was expecting to see a higher number of posts in the Spanish category and as I did not get this expected result, I plotted the Number of posts against the number of votes in each category.
From this chart it is easy to see the outliers. Spanish having a slightly higher than average number of posts, but as it is placed well over on the left of this chart, there has been well above average number of votes by witnesses in this category. Photography and life categories are also outliers with a rather high number of posts that were voted on.
Next I looked to see if witnesses tend to vote on posts or on comments. In the table below 0 relates to a post, other number related to the comment depth on a post
I was surprised to see the number of votes given on comments of a depth of 2 or more. It shows that discussion on posts are read by witnesses. So I wanted to see which witnesses were actively voting on comments with a depth of 2 or more. From the chart below we can see a breakdown of these votes, by witness. It is clear to see that some people are spreading comment votes around the community, while others like to vote on their own comments!
Downvotes
The final table I wish to share with you is the witnesses with the highest number of down votes to other people posts
This analysis probably opens up more questions than it answers. Questions that the data can answer, however I have purposely left out some of the questions I asked.
Why?
Well there is a substantial risk of down votes for bringing to the forefront things people don’t want discussed.
I am part of a Steemit Business Intelligence community. We all post under the tag #BIsteemit. If you have an analysis you would like carried out on Steemit data, please do contact me or any of the #bisteemit team and we will do our best to help you...