Self-votes, low-diversity votes and how the situation changed since 2018

5개월 전


Image: modified screenshot from steemit.com

Self-voting and "vote farming" is a hot topic here on Steem. It is no secret, that a couple of accounts get along quite well from self-votes, wealthy friends, or well-staked alt accounts. Some make no secret about it, others try to hide it. There are controversial discussions on how to tackle this and whether it is a problem or not. After all, it is (at least in most cases) the authors very own stake and there are no technical measures in place to avoid rewarding no one but yourself. On the other hand, how should a system like Steem work if we'd all switch to PoW...
I have looked into self voting and low-diversity voting (e.g. voting for only the same small number of authors) a bit more than a year ago here:

Self-votes and Self-Service vs. Spreading the Love - Analyzing Vote Diversity in Steem

The main take-away from April 2018 was:

  • By total vote value, self-votes make up around 6% of all votes.
  • Low-diversity voters (accounts giving 80% or more of their vote value to up to 6 authors) make up 12.8% of all votes by value. Excluding bid-bots, cleaners and dapps, this even makes up a fifth of all remaining vote values!

(How) Has the situation changed since then?

Context: Voters, authors and total vote rshares

voters.png

The graph shows the number of distinct voters and authors per day since the beginning of 2018. As a rough trend, both numbers have mostly decreased throughout 2018 and are currently at around 40k distinct voters per day and around 7k authors per day (incl. root posts and comments).

What may be surprising is that the number of rshares used to vote increases steadily:

total_rshares.png

Included here is also the share of the total rshares coming from bid-bots and dapps. The big dip mid Sept. 2018 was the introduction of HF20 where the chain was unusable for a couple of days.

Self-votes and low-diversity votes

"Finding" self votes in the blockchain data is easy, since it's just checking if the voter is also the author. However, putting the wealthy-friend or alt-accounts into numbers requires a bit more complex analysis. I have selected the following metrics:

  • total value of all self-votes
  • total value of all votes where the voter spends 80% or more of their votes by value to up to 3 accounts
  • total value of all votes where the voter spends 80% or more of their votes by value to up to 5 accounts
  • total value of all votes where the voter spends 80% or more of their votes by value to up to 10 accounts

All 4 metrics are calculated based on the blockchain data from all 1-week periods since January 2018 until July 20th 2019. I have compared the resulting rshares numbers with the total number of rshares of the corresponding week. Here are the results, first in absolute scale:

total_rshares2.png

And the same in a relative scale, compared to the total number of weekly rshares:

total_rshares2_rel.png

The results from April 2018 with around 6% self votes are consistent with the numbers here. What's remarkable is, that the total value of self-votes (and therefore the SP/Vests used to self-vote) has hardly changed throughout 2018! With the overall increasing number of vote rshares per week, their relative share even decreased to around 4% of all votes by value. There's a slight increase since the start of 2019, but I wouldn't have been surprised if the rates increased much more...

Around 7-8% of all votes by value are currently coming from voters that spend 80% or more of their vote value to 3 or fewer authors, 10-12% to 5 or fewer authors. We can take this a step further and compare the self-vote and low-diversity vote values to the total vote value excluding those from dapps and bid-bots to put that more into relation of the "active-user" SP:

total_rshares2_rel.png

This lifts the self-vote rate to around 7-8% of all votes by value and the 3-author voters to 12-13%. The 3, 5, and 10-author voters roughly kept their relative share despite the increase in the number of voting rshares.

Conclusions

  • The absolute value of self-votes (and therefore the SP used to self-vote) remained mostly constant throughout 2018 and is only slightly increasing in 2019
  • An increasing number of total vote rshares made the relative self-vote rate even decrease to currently 4% by value.
  • With the current rshares-to-sbd value, self-votes make up roughly 500 SBD per day
  • The value of votes from voters voting only for the same small number of authors follows the same trend, but is significantly higher with 7-8% for 3-author voters and 10-12% of 5-author-voters.
  • Overall, I would have expected a much higher percentage of self votes and low-diversity votes. Nevertheless, a couple of well-known expected and not-so-expected voters with a low diversity pop up with these metrics.

Are you interested in your own vote distribution? Check

with your username.

Repository

https://github.com/steemit/steem

Tools and Scripts

All data for the analysis has been queried from @arcange's SteemSQL database and fetched, processed and plotted with python and matplotlib. The scripts to reproduce the results and figures are in my GitHub.

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
STEEMKR.COM IS SPONSORED BY
ADVERTISEMENT
Sort Order:  trending

As we are signing off for a little while, I'll give a non-utopian mod comment before the official one and say thank for all the good work you do behind the scenes and with these analyses. I suspect no-one will argue if the reward is as high as the category allows and hope that our reviews in the past year or two have been fair.

Cheers, I'll be back :)

·

Haha, thanks Asher, also for your continuous support - behind the scenes and on chain - highly appreciated! :)

·

Thank you for your review, @abh12345! Keep up the good work!

Nice work @crokkon - resteemed :) I would be curious to see the correlation between bid bot (paid vote) use and self-voting - I would assume it is high, as getting a self-vote in before large paid votes is one of the best curation ROI games in town particularly when you can wait to decide what post to pay for votes on until after you manage to get your own vote in at the optimal time. Any way you could run a quick query using your bid bot list to see what % of posts upvoted by the self-voting and low-diversity voters also end up receiving paid votes, vs. what % of posts NOT upvoted by self/low-diversity voters end up receiving paid votes.

·

Thanks @carlgnash! Oh, very interesting aspects. The order of votes is something I completely left out here. Another aspect - probably going into the same direction - would be to look into the distribution of self-votes between root posts and comments. I would expect a good share of them to be on comments. I believe curation sniping on self-voted bot-boosted posts isn't as easy as it might seem. There are dozens of bots out there specialized on exactly that, and they will be faster than the selfvoter and make better returns after only a few attempts...

·
·

Yeah order of votes is obviously important, I am actually working on an analysis post currently that I hope to publish tonight RE paid votes share of reward pool, and I chose to focus on share of curation reward claimed by bid bots vs. r-shares rewarded precisely because r-shares doesn't reflect curation sniping and vote order. Interesting point RE comments vs. top level, I think it would be worth a quick follow up post from you showing same metrics you have above but split out by top level vs. comments.

I would still expect a much higher % of posts that receive paid votes to have the author's vote on it, than the % of posts that receive paid votes and have no direct self vote. Honestly wouldn't be surprised if it was near 100% of posts that receive paid votes that also have direct self vote. There are social aspects to consider outside of pure curation reward mechanic, the "no bid bot camp" has high degree of overlap with the "no self vote camp" :)

·
·
·

Looking forward to your post! For the curation aspects the order is indeed crucial. Thanks for the thoughts, you got me plenty of ideas for follow-up topics :)
I also wouldn't be surprised if we'd find a strong overlap of the "bid bot camp" and "self vote camp".

Hi @crokkon,

Thank you so much for your great post as always! Probably this might be your last utopian analysis post? I'll miss your post.

Probably it's good idea for you and @abh12345 and other moderators to apply for steemit delegation with a program focused on steemit github. only steemit github and put much more weight on analysis, dev than other categroies.

·

Hey @blockchainstudio, yeah, the Utopian changes came rather unexpected for me. I still have a couple of ideas for posts, but you know best the time it takes to work out analysis topics :) It might indeed be my last utopian post before operations stop end of the month. Interesting idea with the delegation program - but for that I'd await more information on Utopian's future. Thanks for your comment!

·
·

Thanks, hope there'll be a similar program.

·

Last for a little while maybe, but hopefully there will be more in the future!

I hope the top bods at utopian are aware of the steemit delegation post and will be discussing it :)

·
·

good to hear that. there should be.

Please note that selling a vote is pretty much equivalent of a self-vote (as in this scenario, one get paid for a vote from author rewards regardless of content).

Alice has three accounts: x, y, z,
x voting for y
y voting for z
z voting for x
No self voting? Right?
Wrong.

:-)

Hi @crokkon

Thanks for this excellent contribution which i've given as a 'staff pick' this week :)

Like you, I'm very surprised about the stability of these numbers over the past year or two. I think many had thought that the game was up and most had resorted to self-voting across 1-5 accounts but the results so this isn't the case.

I think the issue is amplified in that some of the self / low account spread voters are large accounts which have been pretty much left to their own devices for a long time now.

Will the next fork help in any way? I hope so and wonder if over valued / botted posts could be targeted first, then obvious self-voters, and then if there is downvote VP left to go around, people may want to see who is farming other accounts with their vote.

Great post showing things maybe aren't as bad as they seem :)

Cheers

Asher

Your contribution has been evaluated according to Utopian policies and guidelines, as well as a predefined set of questions pertaining to the category.

To view those questions and the relevant answers related to your post, click here.


Need help? Chat with us on Discord.

[utopian-moderator]

·

Hi Asher, thanks a lot for the review, the resteem and the staff pick :D
Yes, I also perceived the situation as worse compared to what the analysis gave. I think we're easily biased by seeing heavy self-voting in one or the other form from a few well-known accounts. However, the analysis looks into the vote values, not the numbers. With many votes, the relative share of self-/low-div. votes may be higher. I'm not sure if the next fork really helps here. As a self-voting author, I'd technically still get the same rewards (excl. maybe curation snipers), just differently distributed across author and curator share. The HF may, however, make it less attractive to delegate to bots.
Cheers!

My data in the last 31 days: upvoted 264 accounts, 25.1 % are selfvotes. I needed selfvotes to get enough liquid Steem/SBD for contest prizes and donations but now with earning multiple tokens from Steem Engine I think I'll be OK with lesser amount of selfvotes. What do you think?

·

The parallel earning opportunities on SteemEngine are an interesting aspect in this regard - thanks for sharing this! This indeed brings some liquid tokens in addition to STEEM/SBD/SP. The "right" balance on self-votes is in my opinion highly subjective. Technically, everybody has the right to do with their stake whatever they want and I don't believe that we'll ever come to a conclusion on what is accepted from the whole community.

This is great data analysis. Thanks a lot for sharing the tools you used!

Posted using Partiko iOS

Uhmmm maybe a model approach such as the one that palnet took could help to incentivize people voting on other's posts.

·

true. SteemEngine provides a lot of options to tune author/curator shares and reward curves. And from what I understand, HF21 plans to address changes to both parameters as well. Thanks for your comment!

Hmmm interesting findings and if the vast majority of users aren't self upvoting then we have a load of legends using the platform. It's not as bad as some have made it out to be! Thanks for the data crunching!

Posted using Partiko Android

·

Be careful: this is by vote value, not by the number of votes. While there are around 4% self-votes by their total value, there may be significantly more by numbers or users esp. if smaller accounts self-vote. But yes, I also would have expected the situation to be worse.

·
·

Let me enjoy the blissful ignorance for just a moment... enjoys moment before realisation that he slightly misunderstood the research

Nooooooooooo haha

Posted using Partiko Android

Hello!

This post has been manually curated, resteemed
and gifted with some virtually delicious cake
from the @helpiecake curation team!

Much love to you from all of us at @helpie!
Keep up the great work!


helpiecake

Manually curated by @niallon11.


@helpie is a Community Witness.
For more information about our project,
please visit this month’s UPDATE post.

·

Thanks for bringing some cake, @niallon11! :)


This post was shared in the Curation Collective Discord community for curators, and upvoted and resteemed by the @c-squared community account after manual review.
@c-squared runs a community witness. Please consider using one of your witness votes on us here

Hey, @crokkon!

Thanks for contributing on Utopian.
Congratulations! Your contribution was Staff Picked to receive a maximum vote for the analysis category on Utopian for being of significant value to the project and the open source community.

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!

Hello, as a member of @steemdunk you have received a free courtesy boost! Steemdunk is an automated curation platform that is easy to use and built for the community. Join us at https://steemdunk.xyz

Upvote this comment to support the bot and increase your future rewards!

Hi @crokkon, excellent contribution :)