Punishing Rabid Self-Upvoting of Comments
So . . . . with HF19, which overall was awesome, we have a new, and apparently reasonably major, threat to the community. As pointed out by Jerry Banfield last week, We Double Our Steem Power Upvoting Ourselves Every 181 Days!
Now, at first, this may appear to be a minor problem rather than a major threat -- but let me lay the statistics on you. According to SteemDB, in the twelve days since HF19, rewards on comments have risen from 77 MVests per 30 days to 322 MVests per 30 days. Correcting for the fact that the majority of those 30 days is still pre-HF19 shows that the expected 30-day *increase* will be *at least* (((322 - 77) / 12) * 30) = 612.5 MVests.
Rewards on comments have increased by nearly an order of magnitude
If this still doesn't seem to be a problem, consider that 50MVests is approximately 1% of the distribution of new Steem per 30 days. So that 612 MVest increase means . . . .
Over 12% of new Steem distribution has newly been diverted to rewarding comments -- which means that author and curation rewards are *dropping* by a similar factor.
And this all assumes that the problem isn't still getting worse -- which is contradicted by the fact post values still seem to be dropping with time (an indication that the average rate of votes markedly increased after the post was posted).
Here are some of the worst offenders (all drawn from the 1000 largest accounts). Data was drawn from witness @jesta's awesome SteemDB.
There are several ways to solve this problem.
One is to pray to the programmers and ask for divine intervention in another hard-fork. Another way is to handle it as a community with altruistic punishment. If a egregious self-promoter gets their comments flagged after they up-vote them, they won't derive any benefit from doing so. If they persist after being warned, a post or two of theirs can be flagged as punishment. That should stop the bad behavior.
This account was originally intended to be a community-governed bot. I intend to start moving it towards that purpose now. Until the community votes otherwise, this account's voting power (except for the 500SP already delegated to @MinnowSupportProject) will be used *solely* to try to stop this problem. If you want to support this effort, you can help by temporarily delegating some Steem Power to the effort (or, of course, by up-voting this account so retaliatory flags don't drive it into useless oblivion).
@timcliff has created an excellent tutorial on how to use @busy.org's tool to delegate SP. The single link version to delegate 100SP is https://v2.steemconnect.com/sign/delegateVestingShares?delegator=<your-account>&delegatee=ethical-ai&vesting_shares=206940.794200%20VESTS where you need to replace <your-account> with your account name and the 206940.794200 with whatever multiple changes the amount from 100SP to your desired amount (important: it must maintain 6 decimal places).
Tomorrow morning, I will start manually flagging (with an explanatory comment) egregious new occurrences of self-up-voting comments starting with the offenders in the chart above.
I will also post a list of those flagged so that others may join in as desired.
Hopefully, together, we can bring this scourge under control . . . .
Major props to @aggroed, one of the founders of the Minnow Support Project, for calling this problem to my attention.
Then the HF was not awesome.
Full linear rewards + 4x vote power = Much larger potential for abuse/exploitation
Neither of these changes were necessary. Full linear rewards, in and of itself, represented a serious issue for mitigating self-voting and sock puppets/collusive voting. Add the ability to give a 4x weighted vote, and it presents an environment that's ripe for "draining" the reward pool on spam and other useless/worthless content.
We've all seen what has happened over the last 12 days. I think it's safe to say that the changes have so far reversed some of the "social" aspects of the platform and have created a much more lucrative opportunity for spammers and the "get-rich-quick" users who have no interest in any long-term viability. It has simply sparked a furious cash grab and the reward pool fund and general behavior of users have demonstrated that.
This HF should be rolled back, as soon as possible.
I agree with the majority of the witnesses that the hard-fork was too much, too fast.
On the other hand, n-squared rewards was abusive and went on FAR too long -- so rolling back the change would be a really detrimental idea. It would have been far wiser for the developers to have suggested something between the two (and I believe that collusive voting between whales was reduced by the change far more than the change increased the sock-puppet/collusive problem).
The 4x weighted vote should have only been given to the smaller users and should have been part of a separate fork.
There is nothing "abusive" about n^2, but it was far too much in terms of calculating post payouts. Pretty much everyone agreed that n^2 needed to change to a more linear algorithm. But there wasn't much agreement on a full linear algorithm, and self-voting/sock puppet voting was one of the reasons why. Many people spoke out against this change precisely because of the behavior we're seeing now.
With a more-than-linear curve, it would be much harder to give yourself large rewards without other voters agreeing on rewarding the content. Obviously, full linear requires no such agreement from other users. So having a more-than-linear algorithm is at least a moderate deterrent to naked self-voting for rewards.
But the 10-vote target was a "solution" for a problem that did not exist. And rather than make things more "fair" or "equal" for smaller stakeholders, it simply allows spammers or anyone else to siphon rewards into their own pockets with minimal content that requires not even the slightest bit of consensus...or any agreement whatsoever from even one other user.
It's the combination of the two changes that really exacerbates the effects/problems. Even if they were implemented individually, either one would still create similar problems. That's why I say they should be rolled back.
Change the full linear algorithm to less than squared but more than linear. (STINC has claimed this is too hard to do.)
Get rid of the 10-vote target altogether. It was completely unnecessary.
10 x more blogs each day this past month prior to the first 4/5 months of the year and a '10 vote target'.....
Bog it, i'll just vote for myself mentality!
This comment is also for @ethical-ai and all the others commenting here on the post. I place this comment here for visibility.
I do not agree the HF should be rolled back. Next HF may need to be adjusted.
Disclaimer: I started writing this comment having in mind to list only a few items and not more than 100 words, but it became a really large comment; Since I went into depth here and there, doesn't mean I expressed all the details I would express when we would have a long debate; I may also not have mentioned some exceptions to the rule. And I certainly did not mention all the things that are 'wrong' - in my opinion in this community - wrt who gets votes and who not; for sure a fact is that some of us get enormous rewards given by high SP holders ALL the time and ending up in Trending ALL the time and many good authors and posts do hardly get any votes/rewards at all - just to name another topic of total disproportional imbalance.
Long comment with valid points.
I always liked the idea of "social exposure" with warnings first, to have a chance to change the so called "bad" behavior.
Good comment and valid - enjoyed reading the whole thing - even though it was long :)
Until a few hours ago I thought HF19 was a success - now it seems not so much now.
It's a shame there isn't a way to test these changes before going live with new code. I'd agree and say voting culture is at the core of this.
Testing is not really possible, other then try in the field and decide next HF to implement something else or step back to what we had.
Flaging people for upvoting their comments is not solution. Why then we don't flag anyone who upvotes their own post? Or anyone who uses bots?
Upvoting a post or a comment that is well-thought out and valuable is not the problem. The problem is people who are spamming Steemit with dozens/day simply so that they can upvote them.
Bots should be flagged if they are performing anti-community actions and supported if they are performing pro-community actions (i.e. cheetah, etc.)
You can't solve problem of spammers. Only thing you can do is to unfollow them.
Unfollowing them simply leaves you blind to what's going on, essentially ignoring the problem. It needs to be fixed, not ignored.
HF20 disabling upvoting your comments and posts...solution is simple.
Right up until someone creates a second account to simply upvote their own posts on their main account. I feel the biggest part of the problem is that minnows (including myself) have been given too much influence.
Up-voting your own posts that represent a lot of work is reasonable. Up-voting a half dozen or more small comments PER DAY is not.
So let them put algorithm(bot) in HF20 which will flag account which only upvote same one or two accounts
Maybe someone has only a couple friends on the network, and decide to spend their votes on only those friends. A flagging "bot" could punish people who are doing nothing wrong.
I disagree with self upvoting "SHUN LISTS" a great deal, and here is why.
First of all, do lists like this, REALLY ASSESS the quality of people's comments, they are upvoting for themselves?
Secondly, do they assess if the same up-voters are also spreading the love for others they are interacting with?
Thirdly, should the community be deciding these things as self-formed vigilante groups, or petitioning the developers of steemit, to change the self voting rules?
I could go on and on and on, but I think that pretty well sums up several of my ideas, and I will now self vote myself, as well as several others, that I liked that commented as well.
PS - please let me know if you plan to shun me, so I can be sure not to:
Thanks ;-)
PS -as a fairly new member of steemit, I spent time asking people if it was ok if I upvoted my own posts. Before HF19, people told me it was perfectly acceptable, so I find this a new issue for our community. Not a tired old issue..
I would respectfully suggest, if we are concerned people are not contributing to the discussions, then we could:
-suggest the same person only gets to upvote one comment per post?
I noticed as a new member of steemit, trying to experiment with what generates SBD in general (which everyone will naturally do by the way):
Finally, such "witch hunt lists" as above:
correct?
I don't think they know what they are talking about. There are many real issues to focus, what they bring is non-issue. If self upvote in any form is really an issue, free market will fix it. It is none of their business, everybody should run their blog the way they want. That is the point of having stakes with SP. These guys are just hating and trying to establish authocracy lol. They cry about self upvote, but don't have decency to upvote those who engage with them.
I would like to point out:
I have been trying to learn more about how thing work as well. At this point my understanding is top 20 witnesses decide what changes will be applied in hard forks. However everybody gets to vote for witnesses. Ultimately they are the ones who make changes. High SP holders are already in loss because they don't get 90-100% interest on their SP as it was originally. It is very small now. Why would anybody tigh up their steem in SP, if they cannot use on themselves as well. So, this is a failed campaign. In my opinion it goes against basic principles of steem.
well said!
obvious!
couldn't agree more!
what motivation to invest?!
it's the other side of the coin, it must b considered very seriously...
Rather than flag - make the effort to change the formula to more accurately differentiate between comments and posts; and self-upvoting versus 2nd party upvotes.
In the long run the formula will be more effective.
I would argue that always relying on hard-forks (and worse yet, attempting to perfect formulas in a changing world) is a bad idea.
I've seen you say that relying on a hard fork is bad, and in a general sense I agree. But the point behind this platform is that healthy behavior is incentivized.
Saving all your voting power for yourself is not healthy for the platform, yet it is the best strategy. That means that unhealthy behavior is incentivized by the platform itself. In these cases, the platform should change. It's not even much of a grey area in my mind. If whales are trying to retain their steem value, it should at least be in a method that is social.
Other issues, such as copyrighted content can be regulated by the community.
And to people that say you can't stop self-voting because the offenders will just create a new account that always votes their main account.... that can be stopped as well. Just make repeated votes for the same account have diminishing returns. Make it so that you can vote for your friends, but only so often until your power for them recharges. At some point you're incentivized to spread the votes around.
Without this change I see steemit becoming a platform of mass hording.
I'm not sure changes on self voting algo MUST BE DONE in order for the platform to succeed.
This issue could be highlighted as something we revere as meaningful (in the FAQ) and best practice reviewed, and left to (for real people) user's discretion; (find bots that offend and wipe account); and everyone should consider further over time.
That may do it.
If not, then yes, algos need to change. As always.
You're right; it may not even be a question once considered seriously; and it may be on the next HF or two...
But that brings up another question; how much testing and consideration should an issue receive, before it is included in an HF?
Only your witnesses know; but not sure they were entirely in control of HF19 either..
@mark-waser no argument needed "always relying" is the point we agree on however if a formula or feature is flawed then it must be addressed via program code
MARK! Long time no see, my friend! Glad to run into you. And yes, totally agree....but I also believe that creating a context from which "ethical" or " good" behavior can bloom from is a step forward. Of course, ultimately the majority of users will be responsible for what Steemit will look like but let's not give those "good" majority an impossible task.
@razvanelulmarin indeed - we should do what we can to incentivize good and helpful activities.
I certainly do not imply "all bad behaviour" can be solved with algos.
Wow! Where are you at/what are you doing these days? Are you going to Lisbon?
I agree with this if it can be done
The ability to self upvote could also be removed... I'm just offering the suggestion for physical solutions.
At the end of the day you can't remove all cheaters, by changing the algorithm. In fact it would be contra-indicative. Having accounts manage this issue at this time is the right level. If we keep changing the rules to handle every case of cheating, it will create an extremely fragile system. In fact the more rules added, the easier to cheat and harder to identify cheating.
@gutzofter we agree that all cheaters and "gaming" can't be eradicated by the algos.
Personally, I like the idea of limited self-upvotes...
There is a flaw in your description of problem. Comment upvotes are not necessarily only self upvotes. When topic is being discussed some like to upvote the person who is engaging and also upvite themselves, because the other person doesn't have enough sp to upvote the upvoter. What is wrong with sharing upvote 50% / 50% ? I know at least one person on that least does 50/50. I think that is brilliant.
No, comment upvotes are not necessarily only self-upvotes. But the numbers for the dirty dozen above ARE all self-upvotes (because I didn't include their votes for the comments of others). PLUS, I'm only talking about punishing self-upvotes (and only the most egregious repeat offenders).
I follow one user on the list, and interact a lot. Whenever he upvotes himself in a comment he upvotes the other person as well. This makes engagements in hist post incentivised. This I bet most don't do, you probably don't do that either. For this reason I think this list is flawed, and argument has no merit.
Furthermore, if anybody tries to punish this behaviour, must me out of his mind. Because there is no way you can be objective and can implemet universally. What happened to decentralization.
The real issue is auto-upvotes by lazy curators. That's what should be penalized. But nobody will do that, because they are whales. If you can't take on whales, leave the little guys alone.
My friendly opinion :)
So . . . . only 3 of the 10 people on the list upvote other people's comments as frequently as their own -- and in all three cases, it looks like virtually exactly one-to-one.
It's also somewhat incredible that you would believe that over 100 self-upvoted comments in 10 days (the stats only cover the first 10) is reasonable.
And you lose your bet.
All I am saying self upvote on comments is good and healthy if it is also shared with engagers. Its part of the platform vision. This should actually be encouraged.
Only self upvote with no regard to engagers is I agree bs. But people will stop engaging with those users anyway, they lose in a long run. In either case these matters should be left alone for free market to decide. Having organized penalty would not be healthy.
How did I lose a bet? I didn't see any incentivised engagement here yet. lol
People on that list would have showed some love already.
Respect :)
Altruistic punishment is part of a truly free market. Self-organized penalties are part of a free market.
You said "I bet" followed by an incorrect statement. I therefore said "You lost your bet"
How is that incorrect statement? I guess you misunderstood my statement. It seems we are talking but not listening. But anyway, good luck with that.
I believe that we follow the same one user and your post makes sense to me.
I had an idea; how about a maximum number of self upvotes per day? Say 2-5 upvotes.
That way you can upvote your posts (which I believe you should be able to do) and then in order to maximise your curation reward you will be incentivised to vote on others content.
Just a thought, I'm still learning how this platform works. :-D
I could live with 5 I guess. Usually I upvote my own, because they are the best on the thread...
2 - 5 self-upvotes is a good suggestion -- always relying on a hard-fork to solve problems isn't
(and self-upvotes on a thoughtful post for visibility is NOT something I want to discourage -- as I have done it myself to get noticed on posts with pages and pages of comments)
follow me please i did it thanks alot!!....
good work
I like this. It's logical and fair.
Upvoted with full power!
Thank you very much @thedamus :-D
I think the solution would be to increase curating rewards to the point that it gave more incentive to use upvotes on curating the content of others, as opposed to using all your upvotes on your own comments.
Curating awards are already 25% which is already too much incentive for bots . . . .
Right now it seems very is very little incentive, for others, to upvote my comments..
How to fix that?
Upvote the comments of others and count on them to reciprocate. It worked here.
I don't think we need a code change but It's definitely nice to know who contributing to the platform and who isn't. Clearly in my mind voting only for yourself means you're not contributing as much as you should. If you can spend the time typing out a comment on someone else's post you can easily read a couple of other comments and give them a like too.
I'm curious if you can post the users with the most rewards and the fewest comments per post. That's something that bugs me if people are getting huge rewards but never talk to the community.
So someone that posts every day, but also upvotes there own comments, is OK, but not if they don't post daily?
huh, what are saying?
Good post and food for thought but gawd, why do I always feel like I'm part of a witch hunt around here? I've been known to occasionally vote my comments up but guess I'm going to have to rethink that behavior; I'm surprised I didn't make your list.
I'm starting to daydream about men with harpoons chasing me around yelling, "Let's get that crazy dolphin he voted for himself!", " Yah! His blog sucks, it's definitely written by the Anti-Steemit!", "He's corrupting the youth, make him drink hemlock!"
"Dudes, dolphins are not for harpooning, they wind up in the tuna can once in awhile but not by choice. The whale went that a way... ", daydream ends.
Anyhow, other coins are a bit easier to invest in. You just buy your way onto the little piggies list and mint coins on your computer and cash in. In my opinion we should try make it less tough for the big investors to work the system in particular if they don't have the skills, time or willingness to properly blog.
I'm not saying that greed is a proper way to build a community; voting for others is really in everyone's best long term interest. However, many people are making big time cash rewards without purchasing any Steem or putting in any cash (crypto) investment. If you are able to make any significant value from your votes you had to have invested or at least posted something of value. Without investment the price will tank and you can kiss goodbye getting any significant reward at all for your posts.
hey...I'll upvote you if you upvote me...
Just joking...no really :) (I think we need a little comedy here)
Okay, tag your it! :) LOL. It is a bit of a funny situation if you think about it.
yeah - exactly :)
Witch hunt, well said!
I've made several comments above, you can check..
You saw, @l0k1 was was testing his smackdown kitty today? It seems that his bot can be a good solution for "self-upvote" problem.
No! I missed that. I should have a chart with him! Thank you! (+1 from my alter-ego coming shortly)