A Novel Idea For Curation - Add Comment Rank As Coefficient, Kill 30 (15) Minute Timer
Curation remains one of those intractable problems on Steemit. It's a critical activity for the platform, given it determines what articles show up in "Hot" and "Trending", and lack of enough quality curation causes no end of consternation. One can hardly throw a proverbial stone without hitting an argument about some knock-on effect from the community's attempts to solve curation, such as bid-bots and other voting bots.
The difficulty has always stemmed largely from the fact that any curation formula must be expressed mathematically, and this gives voting bots a distinct game-theoretical advantage that forces humans to use them or be left behind.
In the past, I've suggested admittedly convoluted ideas because there seemed no simple solution. Adding a second reputation coefficient, a "Curation Reputation" if you will, based on how many reputation ranks authors increase from others after you vote on their posts (or other factors, such as the current payout scheme) was one. However, these fail the "easy explanation" test and mean making a complex system even more obfuscated for the new user.
I have a possible solution, though there may be many avenues of abuse.
What if curation rewards were modified by a coefficient of how highly rated your comment on the post in question is? Rewards reduced from some users in this fashion would cover the increase in rewards for active, quality commenters. Also, while we're at it, let's kill off the 30 minute (15 in HF20) curation reverse auction. Timers aren't doing much to dissuade bots - they speak math as a first language.
Consider these example numbers.
VaderBot(tm) and Luke (a real person) both 100% vote on a post at the same time, with a 10,000 SP vote weight. Suppose that once the post pays rewards, each are entitled to a total of $1 in curation rewards. Let's presume that they are the only two voters (I know curation doesn't actually pay this well, that's the problem!)
Under the current scheme, both Luke, our hypothetical organic protagonist, and VaderBot, our voting curation bot, would receive the same rewards. However, with the simple addition of a multiplicative coefficient to the final curation rewards calculation, this is no longer true.
Let our coefficient, X, = 50% (.5). This represents the maximum curation adjustment for either not commenting, or having the "bottom" trending comment.
Back to our hypothetical post, suppose Luke posts a comment, which the author upvotes to the top. VaderBot posts no comment. Under my hypothetical new formula, this is how curation rewards would change for each:
Obviously, these numbers could be adjusted to whatever we want. A 0.25% coefficient would render bots highly disadvantaged, which for voting, I'm not sure is negative. Rewards and penalties can scale across the comment string by percentiles.
This clearly gives a huge incentive to create quality commenting and discussion, which is sorely needed in many Steemit blog posts.
The obvious problems with this system will be the ability to manipulate comment payouts to earn curation, however this may be a surmountable problem. Self-votes could be removed from the calculation for "trending" rank (not in general, specifically for curation rewards) if the community feels that is a valid adjustment, which would prevent whales from entering posts late and voting themselves curation. The (admittedly relatively small) share of rewards that go to curation may not be enough to cause users to risk their reputation by modifying their behavior negatively in such a fashion to begin with.
Vote trading is another issue much more difficult to prevent. Voting Bots that currently allow you to use them on comments are a serious issue. They will create an arms race for curation rewards, but they are already abused now on nonsense comments so this may simply result in these abusers being flagged or un-voted by bot operators more quickly. A concerted effort by the owners of bots to un-vote early abuse without refund, while keeping a community blacklist, could quickly render this activity economically -EV. Or, those bots who allow comments can simply disable that function, although this would likely create economic pressure for a black-market bid-bot without such a restriction.
I'm not suggesting this change be made, but it seems worth considering.
We also have a Radio Station! (click me)
...and a 5000+ active user Discord Chat Server! (click me)
*Sources: Ebay, Google, *
Copyright: Star Wars, Penguin, Random House, ALittlePinchOfPerfect.com
This raise the question do we want to have bots vote for our content? A lot of web side have anti bots mechanics, but because of how the blockchain works this is not feasible, a test for IQ is a way to filter curators opinions and not the content and this not guaranteed for good content anyways, I think the root of the problem is the lack of people in Steemit, and lack of interest to update the site.
"but because of how the blockchain works this is not feasible"
Correct, we can't make bots technically impossible, but we can disincentivize their use by rewarding real interaction. This doesn't mean we want to eliminate all bots, but it is a tool for regulating what proportion of blockchain interaction are automated.
It's a lot harder to write a bot that posts a salient comment than simply votes.
You are right, bots don't have that capabilities for doing that "yet", in the other words, we will making people develop smartest bots each day by making harder for dumb bots to make money, this is a good and easy way to promote AI development and I support this idea.
It is interesting to think of this as simply an AI arms race to producing the fully human-impersonating Steemit robot.
I think users that intelligently interact with other users will outperform bots in the long run. I'm pretty sure that @m3mt and @lexiconical are humans. That is why I follow them and would up-vote but can't afford the power right now.
Bots and humans that do not contribute value to a post will get a down vote if I could afford it. I'm looking forward to being able to divide my up-vote/down-vote so it don't deplete so fast. This feature should be more readily available for more users so that that non-valuable posts/comments get down-voted more.
Grammar is also very important. I am rarely going to follow/up-vote/reply to users that don't use proper spelling, punctuation and grammar - they should at least put some effort into it.
"Bots and humans that do not contribute value to a post will get a down vote"
Only if they are posting a target which can be flagged...
"Grammar is also very important."
I wish this were more generally true.
I'm going to have to disagree with you there, I'm pretty sure @lexiconical is at least a Vulcan or maybe a Metron.
I absolutely never hang out here:
https://www.reddit.com/r/totallynotrobots/
Eventually humanity will grow obsolete, even if it is hard to believe, for now I think it is a a little of a wasting of our time to make it "harder", a temporarily fix it is just that, it would be admitting you can't beat bots no matter what, if you know the end result then you will be only wasting time in solutions that only create more problems.
But what if we get rid of the problem altogether? What if we make absolutely sure one real person is casting one real vote and he or she is accountable to his own vote? Haha that would be crazy expensive right? But it would get rid of all complains for complete, we know the answer and nevertheless, who would be willing?
I like practical solutions, I made my tesis base on genetics algorithms that give a "good" solutions but not the best, so yeah, I agreed with you, we should make harder for bots to vote and give some breathing room for humans, now I leave you with this question, how we make that happen? I think this post is good start.
"humanity will grow obsolete"
By what metric...? I doubt we'll ever see humanity as obsolete.
This bring out two questions, "what is humanity" and "how can be obsolete", I refer to the "human mind and body" as growing obsolete compare to the "mind" and "body" of a robot in terms of durability and how much accurate information they can hold without forgetting anything at all.
I don't know if you are referring in your life time or into the distant future, but I believe that give it enough time and resources, and you will end with an A.I robot that can fool any human into thinking they are talking to another human.
Btw, if you can have an engaging conversation con with a bot, does it matter if it is a bot and not a human? (This is outside Steemit realm anyway it is just a side question)
Either way, the bots are running wild in Steemit, it would be nice to have a test to check if someone is a bot before commenting, flagging and voting, but I doubt it would put in place though, as "it is a big business".
From the perspective of physical capabilities, our obsolescence may already be at hand, but what becomes the purpose of AI if it has nothing to serve?
"Btw, if you can have an engaging conversation con with a bot, does it matter if it is a bot and not a human?"
Perhaps not to you, but it is relevant for the other party in the conversation, I suppose - if they have no need of human contact, one would expect them not to value the process.
"Either way, the bots are running wild in Steemit, it would be nice to have a test to check if someone is a bot before commenting, flagging and voting,"
Yeah, this would probably take too many resources to be practical, as you suggest.
Reading your comment, clicking on your profile, and seeing "FreeLance", made me think again, that we should make a game about "Free Lancing". Something perhaps like flappy birds/joust. A no cost ["Free"] "Lancing" [jousting] game lol. This is partially because you actually have "Free Lance Gamer" Lol :) Alx
You really took this comment in an unexpected direction.
That is generally what I do sir. Once I had enough real ideas to ensure I could make my fortune, I just let my mind fly wildly, and sat back taking notes. I should be publishing all 420 in 2018, first on Steemit and DLive. When I’ll start using the account @opensourcery. :)
You sound like an awfully interesting fellow!
I'm experimenting with a similar process myself in 2018...
hahaha that's an interesting concept for a game, can we put some strategic layer? Btw the only reason why the gamer tag is there is because i like to play a ton of xcom2.
It sounds like more of a gimmicky name idea than a real innovative mechanic, tbh.
Lol I always or usually start out with gimmicky names and then sometimes an idea forms. The best I have is free for all “free lancing” where everyone is trying to impale everyone else on a lance, maybe pubg style [which is a game I “invented” [part of] years ago, on some level, suggesting that we use the huge screens at football fields to show a group of competitors on the field with an overhead camera, that moved around, eliminating any people who couldn’t stay in “frame”, but this was obviously real people, running and jostling, not Video] the idea for free lance is I guess is like Greek triremes, where everyone is vulnerable from the side, but deadly if they strike anyone with the “tip”. Maybe that’s a better game. Don’t trireme, DOREME. Lol.
We could. But I’m sure we have a great many other things to do. Or not. I’m usually at https://discord.gg/2FwzQu talking about all the ideas.
Maybe we could just make it so that everyone has to SPEAK their comment into an app, that then transcribes it, presenting it as text, but also having the audio file available for the first seven days at least, so anyone can verify in a moment if it is a human. Sure, bots will catch up in a few years, but we could get a ton done in a few years, if the system were relatively "perfected". Alx
That is an interesting idea for verification, and would be tough to fake, however I fear the bandwidth requirements instantly render it to costly to implement.
Well on Steemit sure. Maybe that’s the basis of steem too, I’m not an expert, but my instinct tells me that steem, for the value it does have that could be leveraged for amazing shit, isn’t gonna be the chain that enables the change we wish to see. Right now, I can’t “be the chain I wish to see”, because it doesn’t exist yet. For now I mostly just work in discord and post on Steemit. Other than two friends of mine, who now work for Steemit inc. I don’t even know when people of any strength or presence to do anything else here, and the money I have, it’s going towards building something else entirely. Alx#3690
This is not true. Here's how it can be done.
It won't be done, because those with the SP to make the change profit from bots.
They won't end bots.
"because those with the SP to make the change profit from"
It all comes back to this in the end. You can fill in the blank with a number of variables.
Ultimately, one feels forced into the old adage...
"When in Rome..."
WB, BTW. I was beginning to wonder if you had become 'he who was formerly known as @lexiconical'.
While there are good reasons for that adage, I am me wherever I am, and will do things according to my principles, rather than whatever the folks I am around do.
As you might suspect, this isn't the most emunerative policy, nor does it guarantee my freedom in some milieus.
agree with getting rid of 30 minute wait, but even rating comments on how well they are liked? popularity factor. everyone would be upvoting the whale comments and commenting on them looking for a bonus. or upvoting friends. there is no fair way to allocate, other than involvement. perhaps we should give credit to those who write comments that spawn discussion?
"but even rating comments on how well they are liked? popularity factor. everyone would be upvoting the whale comments and commenting on them looking for a bonus. or upvoting friends."
I think you missed my point here, or I'm misunderstanding you. The idea is to determine POST curation rewards partially by who makes the "best" comment. Voting for whale comments would be the opposite of "looking for a bonus" - you'd be lowering your comment coefficient for that post by giving the whale more curation rewards. Of course, this does institute a "prisoner's dilemma" whereby commenters will not want to vote on other commenters due to competing for the same curation...
"perhaps we should give credit to those who write comments that spawn discussion?"
This is precisely what I'm suggesting, although the admittedly imperfect way to judge that is pending payout of the comment. I'm open to suggestions. We can't do something like number of replies, because of bot-spam, but we could do total pending payout of all replies (child-comments to that comment)?
ahhh. ok i misunderstood. i think we can agree on this but... how to easily judge best comment? who decides?
We want to keep it simple. If "pending comment payout, minus self-vote amount" isn't good enough, what else can we use?
I mentioned total payout of child comments, but many threads have no child comments and they rarely have many payouts...
it seems if we use this, we might need a few disqualifiers like, "great article, keep up the good work"
Comments like this should get down-voted.
I started downvoting them regularly when I stopped caring about my blog, reputation level, flags, etc.
It was kind of a relief to be released from the social pressure of conformity.
This should be encouraged - its' one of the greatest strengths that humans have over bots in this environment. I would also like to be able to more easily down-vote foes. When I say foes, I am referring to spammers. To me, spammers are those that post content with little or no added value.
Agreed on all points. This is why I've suggested solutions in the past repeatedly, such as not requiring flags to use the same voting power as upvotes. Instead, we continue to force people to pay to regulate spam...gee, that couldn't go wrong.
What if the bots apart from voting also make simple comments and receive massive upvotes from other bots?
I believe you are right when saying this would create an “arms race for curation rewards”. A flagging war between supporters and people against this.
So I am not quite sure if this could improve steemit.
"What if the bots apart from voting also make simple comments and receive massive upvotes from other bots?"
This wouldn't make economic sense. The bots would be better off self-voting or vote-trading this vote power more secretly. If they engaged in this to put themselves at the top of Steemian's blog posts, I think the community would quickly regulate and flag them.
Not much fear of being retaliated upon for flagging a bot!
Social connections can hardly be imitated by bots. It is not hard to distinguish bot from human. If a bot can make intelligent gestures about a post, then they deserve to be in the conversation.
"then they deserve to be in the conversation."
I never disputed this, but they don't need to have the system advantaging them as well.
It is a very difficult problem to solve, any automated solution will always be exploited by a bot, besides there is no interest on the part of the users that change either, most of the users that are in steemit only want to have more better income, and not it matters if it is for a bot or a human the objective is not a quality site, the target is increase the profits while these can be generated, and this is reinforced when users see that an image copied from the internet has $ 200 of profits while that your post, original is worth nothing.
I do not know what solution would be the most effective, what I can say is that avoiding the use of bots, would be a titanic task and you can not win
"the objective is not a quality site"
Even the CEO has said that Steemit.com is not their priority. The truth is I think Steemit may just be a test-bed for their precious SMTs, so I certainly hope they are a smashing success.
Exactly!
Folks that are only interested in maximizing profits should just go mine BTC.
"Folks that are only interested in maximizing profits should just go mine BTC."
Unfortunately, as long as it remains more profitable or easier to "mine" Steem via abuse than to invest in another DPOS coin or mine something via POW, we're going to see this issue.
It would be interesting to see an environment where accounts with no purchased stake actually have 0 vote weight, making spam less profitable outside bots...but I guess you'll still have the problem of trash post bot spam.
I really like your idea as it would further encourage honest interaction on steemit and split the rewards in a fair way. Someone who curates and interacts should definitely get more than someone who's just upvoting something.
We might still see that system being outsmarted somehow but it will definitely be harder to do than right now. I'm overall curious on the changes and improvements we may see on steem within the next months. Until then I keep upvoting as many good comments as possible :)
You'll find you get much higher curation payouts if you vote on comments, as a result of them generally not having (many) votes ahead of yours, and frequently being past the 30 minute timer.
I used to spend a great deal of time encouraging good commenting with large voting rewards...
This post has received a 0.45 % upvote from @buildawhale thanks to: @lexiconical. Send at least 1 SBD to @buildawhale with a post link in the memo field for a portion of the next vote.
To support our daily curation initiative, please vote on my owner, @themarkymark, as a Steem Witness
Well, as a main commenter I embrace your idea, but I am not so sure it will ever be applied. I haven't see anybody giving updates of the new developments or maybe I have missed the right posts...
Search for posts on Hard Forks, specifically, check out HF20. That's the next implementation.
There's nothing there of much interest, IMO, and I agree with your assessment of the likelihood of such a suggestion.
Wow, I didn't expected your reply after so much time. Can I ask how come that you replied to my comment after so long time. How could you even see it :))
Upvoted on minute 44 ? Is there a good time for up-voting ? I heard you should Upvote 5 to 7 blogs per day. Thanks for the info!
Anything later than 30 minutes is sub-optimal, although sometimes less than 30 minutes is better than being right at 30 minutes. It's a needlessly complicated function.
I would say one vote every 2.4 hours. This is how long it takes for power to recover to 100% after a vote from a full charge. This is my goal as soon as I can stop up-voting for a while.
Yep. This comes out to 20 votes per day.
You got a 4.31% upvote from @upme requested by: @lexiconical.
Send at least 1.5 SBD to @upme with a post link in the memo field to receive upvote next round.
To support our activity, please vote for my master @suggeelson, as a [STEEM Witness](https://steemit.com/~witnesses