RE: Voting Abuse and Ineffective Curation: A proposal for blockchain-level change
Currently a very large percentage of voting is done automatically, without the curator taking any effort to evaluate the quality of posts..
The main game of 'curation' in it's current form is to try and upvote the popular authors before others do. With the change to 50/50 rewards, wouldn't this behavior still continue?
There may be some small incentive to actively find good content, but I worry that we will just be paying more money to the users who are not actually spending the time and effort necessary to actually curate, and are just upvoting the same authors because they know that all the other auto-upvotes will follow.
I'm not really against voting bots or auto-upvote trails, and I don't necessarily disagree with your proposal, but I also struggle to see how it is going to solve the root problem of curation. I don't know if throwing more money at it is necessarily going to incentivize better behavior.
I'm hoping you (or someone else) can make the case that it would actually incentivize a behavior change.
It cuts both ways, though.
Clever bots selectively upvote posts knowing the author will self upvote either manually or with bid bots (often 6 days later).
Making curation more profitable relative to self upvoting will undermine this author behaviour and introduce some more ambiguity for bots.
As to consistently great authors, if the quality starts to slip, those manual votes aren't going to arrive, and the curation bots will be pointed elsewhere.
It'll just take a while to lose momentum.
changing to 50/50 may well end up being a disincentive for content creators. Why should someone who places a vote get half when it is the creator that has put the time and effort into it?
I really think the bigger part of the problem is that the Trending page leans more on the monetary value the voting has created than any other measure of quality.
Surely an algorithm could be devised that reduced the monetary influence and factored in number of views, votes and comments on a post. I know there are a lot of spam comments but with the complexity of algorithms that can be devised these days some method of weighting the comments to factor out some or most of the spam should be possible.
The end result would be the more engagement a post gets, the higher up the trending list it would get. No amount of self-voting or vote bots would artificially inflate the result. It would bring the trending page more into line with other social media sites.
Think of it as a partnership. Without votes by SP holders there is no reward for the author (and without SP holders collectively there are no rewards to give to any authors). Conversely, without content there is nothing for SP holders to vote for, and further no content or engagement to drive growth of the platform and STEEM/SP value. And, finally, without engaged voters investing time and effort into curation (the point of this post really), there is no way to determine which are the excellent posts that most deserve to be rewarded.
As @timcliff explained, curation rewards are split between all of the voters, but even in the extreme simple case of a single voter, it is still a partnership and both partners are critical to success. Recognizing the importance of both posters (and commenters) and voters with a more equitable split (and one which better compensates putting in actual effort to curation than the status quo) is good for everyone. I don't know whether 50/50 is exactly the right number but it is probably closer to the right number than the current 85/15.
80/20 or 75/25, even 70/30 may be okay but the 50/50 diminishes the value of the labour the content creator does.
If a content creator deems his/herself too good for that reward layout, someone else will happily step in. And, if that much higher curation payout makes owning SP sufficiently attractive (and it WILL make it more attractive to own SP), then those content creators that opt out will be very disappointed in their decision to do so.
I hear people talk all the time about "what if Steem reaches $10, $100, $1,000... what would these payouts look like then?", well, I don't see Steem reaching those types of prices with the current parameters, but making it far more attractive to own SP (buy and lock away Steem) is certainly a step in the right direction.
People aren't going to be complaining about "only making half of the rewards" as content creators if Steem does a 10X price move based on SP becoming a more valuable commodity.
If a content creator currently makes $100 at today's Steem price (call it $1.50) and today's payout parameters (85/15%), they will be making ( { [100 / 0.85] x 0.5 } x 10 ) $588.24 if Steem 10X in price (to around $15) and the parameters are changed to 50/50%, all else being equal.
you'd make a wonderful fiat publisher as they do their best to drive the proceeds to the writer lower and lower while their intake of proceeds goes higher on the backs of the creators.
What I say is true. If curators make roughly nothing, then how is Steemit any different than a site like YouTube (only content creators earn)? What's the point in holding SP?
If you want Steem to be an attractive investment, make this so called "Steem Power" actually powerful. Right now it's more like "Steem Weakling", if you ask me.
If, on the other hand, you're glad with a flash in the pan, short time window to cash out on your content before this all collapses in on itself, by all means keep curation as being paid in peanuts for holding a boat-load of SP.
You gotta realize that holding a high amount of SP is a fairly risky thing and it's those very people that have taken this huge risk who're making it possible for these content creators to be paid (in anything more than mere pesos) in the first place! Don't bite the hand that feeds you!
Should all these high SP holders get fed up enough and they all start to sell, guess what happens to the author payouts? Don't be too myopic in how you view this.
The curation rewards are not being given to one individual. They are split among the hundreds (or thousands) of users who upvote it.
The problem with the type of algorithm that you proposed (views, comments, etc) is that they can all be easily manipulated by bots.
Often enough, the curation rewards do all go to one individual, if it were a whale who voted first. Bringing the 30 minutes down to 5 might encourage whales to vote manually if they have a better chance of getting in before curation gets donated to the author. The 50/50 rewards however, will be a huge disincentive for good content creators. They would no longer have that 30 minute handicap, they would only have 5 minutes for minnows to get in and give the author some of the curation rewards which would mean we no longer have the underdog handicap where it's better to upvote an underdog (undiscovered but soon to be popular) than a "popular" (already recognised) author.
If you do change the ratio to 50/50, I would argue that the time for curation donation be INCREASED rather than decreased, to increase the incentive of finding hidden gems and DECREASE the incentive of voting the same posts all the time.
I agree that the 50/50 split would be a big hit to content creators. I would be OK with that if I thought it would mean that we were doing a better job aligning upvotes with quality content, but personally I am very skeptical/doubtful that this would happen as a result.
The 30 minuet limit primarily benefits established authors who already have a voting trail following them.
In this comment you say the 30 minute limit primarily benefits established authors. In the next comment you say a 5 minute window would encourage finding undiscovered quality content.
These comments seem to contradict one another.
Not really. One end favors established authors. The other favors undiscovered authors.
A longer "curation donation" window favours established authors who have inevitable support. This incentivises curators to seek undiscovered authors instead because the curation reward won't be donated.
A shorter "curation donation" window favours the curator who can vote on ANY post that was not recognised within the first 5 minutes that it would be a hit. This incentivises curators to vote for @sweetsssj, @timcliff, @blocktrades, @acidyo, and any other established author who is almost guarenteed to make it to the trending page.
That will inevitably make new authors see the game as rigged.
I see your point of view, but I think it oversimplifies things and also doesn’t account for irrational human behavior. The goal of increasing curation rewards (by changing the window from 30 to 5) is to try and incentivize more manual curation. Generally the established authors are the ones favored by bots, so the hope is that with more manual curators, the less established authors will get more attention from stakeholders. Whether that will actually happen or not remains to be seen.
But don't you think the 30 minute limit (or curation donation window as I call it) should benefit the established author? This actually discourages curators who seek curation rewards from voting for the same authors all the time, because once they become established the curation rewards are harder to catch.
Meaning it's more profitable for a curator to find hidden gems, which means new authors have a chance at growing.
Right now the incentives for curation are not very high. The 30 minuet window benefits the author at the expense of the curator. If we want to provide better incentives for curators to find and upvote undiscovered quality content, changing from 30 minuets to 5 will help with that.
I think it will increase incentive to vote for the same authors all the time.
Can you at least explain, where the incentive to find undiscovered quality content comes from?
You will earn higher curation rewards by voting on an undiscovered post than one that already has a high amount of rewards.
Again?
:) At first I was amazed, how many of bigger fishes were also doing that. Now I don't even comment any more. It would take too much time...
I purpose self upvoting on comments to be removed. Let the blockchain decide who's comments are more important.
Nearly every big account holder owns several accounts which can upvote each other ...
I described the concept of 'diminishing returns' which could discourage self-voting, multi account self-voting and circle voting here.
Now that would change the situation a big way. I think.
from your article: "How about if after each vote on a specific account (including ones own account) each further vote on the same account would lead to significantly less curation reward for the voter and less profit for the upvoted account? Thus, when upvoting an account which I had already upvoted before, my voting power would be smaller than in case I upvote an account which I didn't upvote before."
Great idea in diminishing the curation rewards as they continue to self upvote or circle jerk. I think this would be helpful to the platform indeed. What do we do about all this selling of votes and leasing of Steem power? This also is not good for anyone but whales either?
Perfect! Thank you for the link my friend. I will go read this now.
Apparently it's for visibility as I've been told. So if you have lots of money then your comment is more important. So you upvote it to the top of the list so everyone can see it. It doesn't even matter if the rest of the blockchain thinks it's a good comment it "Deserves" to be read.
That's what I've been told
"Just for visibility" is again, something that bigger self-voters mentioned now and than, why did they do that, and continued with the comment due to the article. What you have been told is pretty accurate :) And I 'd also upvote your comment but I delegated most of my SP, so my upvote is not worth even a cent
To prevent bots manipulating the view count is a simple matter, a 3-5 minute page timer, That is short enough time for picture people to get a real view and plenty of time for a medium length post to get a real view. And would cut down on the ability of the bot to jump from page to page. The views are mostly unique visits so a bot viewing the same page every 3-5 minutes would not effect the view count.
The votes are made at the blockchain level. There would be no way to implement a page timer that a bot could not get around by just posting their vote directly to the blockchain.
Took me a long time to read all the comments, I saw further down another reason this would not work, and that is because some places like Busy.org and utopian.io do not use the steemit views, and do not increase the view count on steemit.
Correct.
what more than putting the weight on the monetary rewards that the trending page now reflects being manipulated through circle jerking on the voting? At least their ability to manipulate becomes more diluted than the ease with which so many get on the trending page now for often rather questionable content.
I feel exactly the same way, especially when spending an hour+ or more on a post as I typically do.
In the end, it will encourage larger users to vote for more content and likely to smaller users. This would be a great thing. It will also encourage more people to vote in general. Is it perfect? No, I don't think we will ever see a perfect balance on a platform that provides financial rewards for creating content.
The goal is to encourage manual curation meaning good authors and content providers should be seeing higher rewards. If this happens, the value of Steem will rise via loop of attracting more great authors to the platform and viewers following them with a want to support their favorite authors. We all agree curation isn't working right now? It's not worth it for those with small SP and curiously same applies to whales also it seems.
We have to start asking what those who decide to not join Steemit think and how can we fix it? Main reason probably is that currently we are not rewarding the best content with highest rewards or even trying to find it.
I suspect curation bots will likely continue to upvote popular authors as their main strategy, because I think it's challenging to create a bot that has enough intelligence to do much more. I do think these rules will put human curators on a more even footing when competing for the rewards on such posts. But nowadays the root problem of curation is related to the small number of actively working human curators with enough SP to impact a post's performance and I think this can directly be traced back to the fact that it is not very profitable to curate.
I think once curating becomes more profitable, more users would put effort on the manual curating. But then again, this can also lead to new bots with more complex intelligence.
If Steem incentivized the creation of bots able to understand the quality of posts as well as a human, Steem would become instantly famous in the world.
I sort of don't want this last comment of yours to go unnoticed. It may well be the most important, more so even than your suggestions.
If there were significant incentives for bot creators to compete to make bots more and more effective at content curation, that would be an unimaginable have changer!
steemit
Well, @ned made several delegations of $500k recently, IIRC, and but one of the delegates didn't begin self-voting or selling votes, and kudos to @surpassinggoogle for that.
So, he withdrew the delegations to all but @surpassinggoogle, and is now considering (so I hear) making more, but smaller delegations.
https://steemit.com/voteselling/@stellabelle/dear-freedom-and-whales-here-s-a-sp-delegatee-menu-to-solve-the-vote-selling-problem
[full disclosure: I was nominated for a delegation in that post. This may make you suspect that I am biased. I have not received a delegation, so you can dispel that suspicion.]
Such small delegations might reward a self-voter with up to $5/day, far less incentive to self-vote than the $500/day @ned's original delegations made possible.
One of the recipients of these $5k delegations has voted over 1000 times in the last week (nearly 1200 IIRC). If those votes were worth $.10, then since the median post payout was $.01 for the last month, each of those votes was potentially ten times more than the author might have generally expected from a post.
That such small votes are an order of magnitude larger than most people receive for their entire post is telling that the problem of distribution is terrible for new users.
It also shows that moderate delegations to the stake of many curators can deeply impact krill, and perhaps affect retention.
$5M in such delegations would allow 1k curators to impact up to 10k accounts with 100% upvotes each day, with up to 50 times the author rewards krill receive currently per post.
This is a significant improvement over current distribution, and would be likely to strongly impact user retention.
However, @fulltimegeek and @stellabelle did not make these delegations for a fee. They are perhaps not seeking anything other than to improve the platform. If this effort improves user retention, then that should put upward price pressure on Steem, potentiating capital gains for investors.
I reckon this is an effective and promising experiment.
Glad you are still curating manually those who are on the bottom and not using a curation bot.
Agreed Tim.
Like the account you made for your wife @jerryscamfield where you resteem yourself from and upvote with your bot lol. You're the worst witness this platform has.
Why don't you create multiple witness accounts and buy witness votes for those as well so you can be the greatest Steemian Ever!!!!
Jerry said he hadn't seen the posts and it's possible......He now delivered his promess so my "case" against him is finished
You translated some of scamfields posts into Portuguese and then he didn't pay you like he said he would?
worst...he didn't even upvote it as he said he would...pathological lier...he is a psychopath...lol....he has a pathological problem delivering his word,....it's pathological since this time it wasn't cost him almost anything...just upvoting....his mother should have burnt him with fire when he was little and used to do those things psychos do when they are young like burning cats alive or removing the wings from the flies.
^ This here is exactly what's the problem, not rewarding people for manual curation. However I am kinda curious how this 5min change is gonna play out.
I wonder if having a randomised time after posting, instead of a set 5 (or 30) minutes. In fact, if it was randomly set for each post by the blockchain, but advertised at the bottom of the post, then human curators could decide when it was best for them to upvote, but bots couldn't automatically game the system. That would encourage more human curation over bot curation.
How would the website be able to display a value from the blockchain, without a bot being able to read it too?
Yeah, I thought of that just after I posted. ;). Perhaps it could be in picture form.
But maybe just randomising it between known limits and not advertising it could be enough to disincentive the bots a bit. And maybe not having the author:curator ratio at 1 initially. Maybe 0.8, so the human curators can still get something out of it.
I don't know really. Just chucking ideas around.
I have an idea to chuck out there, the blockchain recognizes the vote, can the vote not be tied into the page view. Human curators I assume go to the page they are voting on or thinking about voting on, so before the blockchain accepts a vote it checks to see if the voter is on the page. If nothing else this will make the page views skyrocket when those 1234 votes need to be on the page for their vote to count.
I ask as an Inexperitse of course. What if an Authenticator code could be under each post to give the voter permission to vote? That is a way to check the real ones from fake voters. Or is it huge mistake? :)
this idea got shot down as the bots do not need it anyway. They interact directly with the chain, not through the interface. From what I have heard, there is no way to identify a bot from human. I keep pushing this point though.
Ive now understood the complexity we are dealing with.
There is a technical challenge to making this happen because of the way the voting system works. When votes are placed, they are done by calling the blockchain API programmatically. The blockchain cannot tell the difference between a human and a bot.
But when you place it there is a door that leaves you place it. Let's assume the Authenticator is the key to open that door. To let you vote. Same problem?
Sorry, I didn't understand. If the blockchain needs to authenticate, it would need to present the token to a UI that would display it to a user. The same token would be presented to a bot. Both a bot and a user would have the same ability to solve it. I don't know how technically you would be able to implement one that woud only work for humans, considering it needs to be implemented at the blockchain level.
I'm understanding it like key encryption. The post delivers a public key to the blockchain, and only having that key allows a vote for the post.
The key is accessible only through the post, via, say, a captcha, which would prevent bots from getting the key, and then voting.
This would then prevent bots from independently transacting with the blockchain, because they wouldn't have the key.
It would work.
Unless my meager understanding missed something important - which wouldn't suprise me =p
got shot down pretty quick by @netuoso -
captcha is easily broken
all it would take is a month or two to get a human driven team getting paid pennies to solve captchas
then you have an even more lucrative system for some hackers since they can get around the block others cant
shitty captcha is broken easily with OCR. quality captcha is outsourced to humans
plus, having to solve a captcha for every upvote would be a huge burden to place on users.
Well, I agree about captchas being totally unwieldy, particularly for every vote.
But, there is a part of the answer, because there is a way to create a portal to the blockchain that necessitates first viewing the post.
Imma think on it more.
Thanks!
That's not even the worst thing. Adding a captcha to a post, which would have to be as some kind of link to a service, introduces a third party service into voting access to a steem post. That is absurd.
Unfortunately, with retention as poor as it is, and distribution being the most likely cause, I am unconvinced it is more absurd than the present situation on Steemit.
I am sure we will have the opportunity to revisit this issue.
Perhaps you can suggest a better means of providing a key on a post which people can access and bots can't?
It's at least an interesting thought exercise.
Great!! then, that.. would provide humongous free advertising & promotion for the Steem token and the Steem blockchain. And definitely, would means that new flourishing faucet based Start-ups & bizniz models would start growing up everywhere like worms all over the internet to earn easy Steem. (same as when the beginnings of bitcoin)
What undoubtedly would attract a bunch of hungry people out there who can't or are unable to write a single phrase as to save their asses, but even so, would contribute to sky-rocketing the popularity and demand of our precious Steem/SBD/SP holdings.
Well, ¡Yeah! at least these new exclusive pauper captcha eaters and prolly also illiterate actors interacting with the Steem blockchain in this way, would be Actual Human Beings milking humbly the system to be able to eat & fill their bellies and not Dumb/Blind/Greedy bot rapists fattenning the already obese asses of the greasy owners of the bots. }:)
¿Burden?, Ha! only for those Lazy Asses sort of users who are not willing to READ or aren't interested on consume good content and/or also refuse to Consciously Work for their rewards.
I mean, I think I've already talked too much. I better shut up now and I start working immediately in the programming and creation of a brand new Steem Faucet Website..... :p
In my view, the cons outweigh the pros - but it is mostly a matter of opinion.
It's an interesting proposal. I'll pass on the idea to some devs to bounce it off a few more people.
Anyway we can implement a captcha on the blockchain where verification is needed before voting?
I am not aware of a way this can be done (technically). If it was on the blockchain, a bot would be able to read it, right?
Well until I saw a comment below reitterating that the bots interact with blockchain and not the UX so this may not be possible.
Quote from tarazkp:
The only other solution I can think of is randomly changing the voting period for max reward per day using voting information from the previous day.
I wouldn't recommend it. However all information on the blockchain is open we can have guidelines for bots.
What if there was a top out threshold that when hit, erased curation rewards and sent all rewards to the author, say $100, that way hoarding and piggybacking on one author could be lessened to some extent, and auto voting would have to be more strategic. It may also help with distribution problems.
You're my witness voting delegate, so I trust you may have a good answer.
The purpose of curation rewards is to incentivize users to vote on the best content. The threshold proposal you suggest would be counter to that goal.
Curation rewards have proven to incentivize gaming curation for rewards.
This is the widely recognized problem that is the reason for this post, and that almost no one will disagree with.
Do you agree that curation rewards incentivize financial manipulation?
I am not convinced that curation rewards incentivize ‘good’ curation.
That appears to be a qualified yes, or at least a probably.
Let me try again.
Are you convinced that curation rewards incentivize 'bad' curation?
Yes, but not 100% bad. I think it incentivizes users to vote for content that they predict will be successful. The original idea was that content that is expected to be successful would be the same thing as content that was actually good. This has turned out to be false in a lot of cases, and it has contributed to the problem of a lot of quality content being overlooked because it is unlikely to get enough attention.
They don't vote on the best though, only the most likely to make the most, some times inspite of lack of content. By capping how much is force funneled into larger accounts, would it not allow for the excess to be spread elsewhere. $100 being arbitrary, it could be $200 or $50
It is not that simple. There are 'good' and 'bad' curators, where 'good' ones are the ones that are actually taking the time to search out and reward good content.
If they are finding it consistently at the top of the trending page, how hard are they truly searching? I have no doubt there are random votes that find there way to the netherlands of steem, but most votes are precast amongst a select group ad infinitum, I'm as guilty as charging in some cases, there isn't always enough hours in a day to seek new, and your favorites will always get the best of you. My idea in no way punishes good content creators other then removing some lazy votes perhaps, it punishes curators who become to complacent in riding the trending page and pop topics like 'steemfest'
The idea of diminishing returns would make it less attractive to upvote the same authors (including oneself) again and again. I described it like that:
"How about if after each vote on a specific account (including ones own account) each further vote on the same account would lead to significantly less curation reward for the voter and less profit for the upvoted account? Thus, when upvoting an account which I had already upvoted before, my voting power would be smaller than in case I upvote an account which I didn't upvote before."
I disagree with your assessment.
This is why I referred that, one of the options could be including proof of work as "READing posts". I mean.. we have already all the tech to make it work. It's just adapt it to the greatest blockchain in question. STEEM it...
Now serious, if people can't really fake the reading and "curation" of something. What better value can we give to it?
Leave you guys to bring me a better reason.
I don’t know if what you describe is technically possible to implement. Also, I don’t think the goal is to eliminate the use of bots. They aren’t necessarily bad in all cases, and they can serve a valuable purpose.
Yes your right about bots. I meant not ending them, but instead ending their possibility to cause bad behaviour, falsifying humanly good curation.
About The idea... the challenge/idea is on the table. I confess it might make little sense to launch ideas that are technically too complex to be implemented in short time or without refactoring the great portion of the code. But my focus was to try seeing or solving problems by allowing more creativity in the options that support human behaviour.
Maybe some one is already experimenting with AI systems and learning with STEEM network, to provide a better bot system.