Recently I was asked on discord to help with some flagging, which I dully undertook after looking at the posts in question.  The story behind the down votes I gave was simple.  The author made a post, upvoted the post, which is fine by me but then went on to leave themselves multiple comments on this post and also upvoted each one of these comments.

There are always people trying to ‘game’ the system and what one person sees and gaming to others it is not.
Then on my Youtube Channel I get this….

![3.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418653/fpx98mepltkf65dypdrb.png)

 
I was very surprised that this comment came to me on YouTube and not here on Steemit.  The account and post that I down voted was 
https://steemit.com/music/@eazyh/izzamuzzic

After receiving this message, my thoughts went to those that often down vote to save the rewards pool.  People like @spaminator, @cheetah and @steemcleaners.  And then I remembered.   I had offered to help @steemcleaners with some data, and never got around to it.

The Aim of this analysis was twofold.

1)	Calculate how much was added back to the rewards pool by these account by down voting and see which account was most active and
2)	Create a model that I could give to @steemcleaners, so they can track, monitor and share data on their work.

The Data
To create the model, I am limiting the data set to Nov for testing and reporting here.
First I connected to the Steemsql database held and managed by @arcange and I connected to the TXvotes tables


       SELECT
      *
        FROM 
        TxVotes (NOLOCK)
       WHERE [voter] in ('steemcleaners','cheetah','spaminator')
       and timestamp >= CONVERT(datetime,'11/01/2017') 
        and timestamp< CONVERT(datetime,'12/01/2017')


I then used a web query to connect to Steemd.com for example https://steemd.com/@cheetah
And here I gathered live data for voting power and effective sp for each account. 
 Each account required a separate query, and then I merged these 3 tables together into one data table.

I also used a web query to connect to https://steemd.com and from here I got other required live data for vote value calculations, such as the exchange rate, total vesting shares and total vesting fund.  All of which is required for calculating how much a vote is worth.

Once I had all the data I needed, I then use DAX to carry out calculations and model the data.
One shortcoming of this model is that in Steemsql, the voter voting power is not recorded, so this has been taken from live data and not the actual rate at time of vote.

To test the data and the model I have used November votes and below are detail of the analysis I carried out

## The Analysis

![4.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418756/jlkcjdji1puwu8b0ezwu.png)

 In total the 3 accounts combined, @steemcleaners, @cheetah and @spaminator have returned $49,259.88 bank to the rewards pool for posts made in November.

2888 different authors received 20820 down votes and the average weight for a down vote used was 3.48%.

The bar chart shows the daily number of down votes and the value of these votes. We can see it was a very busy day on the 17th Nov with $4.2K added back to the rewards pool.

In the pie charts below, on the top left we can see the portion of authors receiving down votes by each of the accounts.  On the top right we can see what portion of the value returned to the rewards pool by each of the accounts. And at the bottom we can see what portion of the total down votes each of the accounts contributed to the over all total.

![5.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418785/buo1c9lor9qamormvldq.png)

 
From this we can clearly see that @cheetah has given 66.7% of the down votes in this analysis but to only 9.82% of the authors.  The level of repeat offenders on the cheetah list is substantially higher than that of @steemcleaners or @spaminator.

@steemcleaners has hit 60.95% of the authors with down vote contributing 95.74% of the value returned to the reward pool.

@Cheetah 
![6.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418802/xikjgotfq1oih84hwn6m.png)

 
@spaminator
![7.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418816/kbgx3mrvrmibhomkmwjv.png)
 

@steemcleaners
![9.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418882/kynf6iroeedxdnifoink.png)

 
##  Repeat offenders

Many of the accounts that received down votes, received more than one down vote.  Below is a list of the authors that received down votes sorted by the number of posts with down votes

![10.png](https://res.cloudinary.com/hpiynhbhq/image/upload/v1513418934/phkeksgwwxosxzdcsyjh.png)

## Conclusion
The data above gives a clear indication of the level of activity when it comes to cleaning up steemit. Kudos for all of the work @steemcleaners.   On average these 3 account in Nov down voted 694 times a day.  That’s like one down vote every 2 minutes. 

By analysing the Nov data I have been able to ensure that I can now contribute in a meaningful way to the support of these services, as I have now passed a copy of this model to steemcleaners.

I am part of a Blockchain Business Intelligence community. We all post under the tag #BlockchainBI. If you have an analysis you would like carried out on Steemit or Blockchain data, please do contact me or any of the #BlockchainBI  team and we will do our best to help you... 

You can find #BlockchainBI  on discord https://discordapp.com/invite/JN7Yv7j


<br /><hr/><em>Posted on <a href="https://utopian.io/utopian-io/@paulag/refilling-the-rewards-pool-when-cheaters-are-in-the-mist">Utopian.io -  Rewarding Open Source Contributors</a></em><hr/>