#project-smackdown - the project reporting weekly on the top self comment voters and flagging the top 20 self voted comments every day.

This is the second and final part of the summary of the first two weeks of the new algorithm, aka phase 2.

In this part

First we will look at what kind of daily return on investment values are most common among comment self voters.

Then we will look at the top 20 users when taking all data for all self voters into account for the two week period of 9th to 22nd August 2017.

Lastly we will then compare this with the top 20 daily lists we reported on in part 1.

Daily ROI values for comment self voters

In the above graph we see a histogram chart for all daily svROI values for all users. This shows us which kind of daily svROI values are most common among self voters.

There is a clear skew to the lower end of svROI, with the most numerous bin (category) of 0.02% to 0.03%. Note that this extrapolates to 0.52% to 0.78% per year.

This also puts the very top end into perspective. There are very few days in which users chose to self vote themselves more than say 0.15% (3.9% per year). And there were only 5 occasions when users self voted themselves more than 0.26% (6.76% per year).

All users, top 20 in summary

Here are the top 20 users order by three different metrics:

1. Two week svROI sum
2. Max daily svROI
3. Average daily svROI

The two week svROI sum is just all the daily svROI values added up. The max similarly is just ordering by who had the highest on any day. The average daily svROI is the two week svROI sum divided by the number of days in which the user self voted on their comments.

1. Two week svROI sum

The sum gives an idea of their total taking. This shows who is objectively gained the most in this period. As it is measured in svROI, it is of course relative to their net Steem Power, so is an indication of how much of their SP backed voting rights they exercised towards themselves.

2. Max daily svROI

Max daily svROI shows who was the most active in comment self voting for any one day during this two week period. So it is an indication of who went all out to gain in one day, but says nothing about their over all pattern.

3. Average daily svROI

Average daily svROI however does tell us about their overall pattern in this period. For the days that they self voted, this shows how much self voting they do by average. You could take it as an indication of their consistency of comment self voting.

Analysis

Three users are in the top 20 for all three metrics: @sanevys, @ika497 and @hjk96 .

Several others had a high max and high average, suggesting when they vote on their own comments, they do so powerfully.

Comparing overall top 20 to daily top 20

The bot looks at the top 20 comment self voters (by svROI) per day and flags accordingly. However let's look at how closely the daily svROI top 20 matches up to the overall top 20.

Half of those users we identified looking day by day also appear on the two week overall top 20. 7 of these gain higher position, i.e. they had even more self votes outside the top 20 on other days, while 3 had less. The other 10 were users that did not appear, and so of course gained higher position.

This shows us that looking at comment self voters day by day gets about half of the overall trend in comment self voting. This is actually better than I thought it would be. Conversely, it shows that half of the users which do the most self voting overall can do it below high amounts and build up return over time.

The top 20 users overall are, in order:

1. @sanevys
2. @davidmccauley
3. @ika497
4. @kj50114
5. @hjk96
6. @sikklel
7. @storcogato
8. @cyberblock
9. @makeithappen
10. @hms818
11. @anforo
12. @riosparada
13. @steemcleaners
14. @gwapology
15. @invsw2016
16. @kiporen212
17. @steemdoge
18. @fisch
19. @redbanana
20. @ehujra

Special mention: users who voted on their own comments every day

These users are ultra consistent in self voting and do it every day:

@hms818
@ehujra
@salva82
@funnyman
@del137
@lee5
@jiahn
@djohan
@captaincanary
@webdeals
@nigelmarkdias

Thoughts on the project

We can see here that there are some very high self voters but also that these deviate significantly from the norm. The higher comment self voters self vote typically from 5 to 13 times the amount of the "normal" self voter, and are about 50 times less numerous. In total during this period we saw 667 people self voting to above 0.0001% svROI *

I think over the course of the last two months we have seen self voting decrease significantly. As I said in the first 4 week retrospective I want us at #project-smackdown to take some credit in encouraging that, though we can't know how much.

I am hereby announcing the end of project smackdown

This will be the last report. More details will follow soon on the full reasons behind this decision, but in terms of the data I have to admit that self voting to extreme levels is not as bad as I initially thought. If you add up the objective value it can still be quite high, but when you adjust for SP it becomes less so.

Against my own hopes for the platform, people do choose to self vote themselves, but it is generally not that much. There are some exceptions and we have gone to considerable lengths to show who you are. I encourage you all to rethink your position on self voting and refrain from it entirely. I still think there needs to be a realignment of incentives to encourage this at a system level, but it is up to us all to campaign for this. If you agree, please do your part.

Thank you for reading our reports and for any support you may have lent us. I also want to thank our challengers and detractors. On a system like Steemit I am happy to say that discussions can happen in a constructive way if most parties agree to be cool 😎 so stay cool guys 😉

Shout out to members of the Steem Coop who have contributed towards this project: founding members @the-ego-is-you and @l0k1 / @elfpsice, without whom the project would not have started, especially @the-ego-is-you who along with me have been here for the whole journey; @transisto for delegating massive amounts of SP, whale and STINC ties and info, and a whale's perspective; @calamus056 who while no longer active debated with me vigorously for many hours to the benefit of the project; @rycharde for extremely smart ideas, blockchain change ideas and data analysis on the ground; and most recently @andybets for the input into the last stages here. You rock guys! 😆

Note, this does not affect the search for an flagging of scammers that @sadkitten is doing. This work will continue. 😼

* footnote: we excluded users for svROI below 0.001%, or for payout value less than $0.001, or for SP less than 1,000