The Curation system needs to incentivise "Genuine Content Mining" - Be honest - Are you Upvoting... or BETTING?
This is a response to @dantheman's recent article about "People Rank" and my ensuing discussion with @smooth, @bendjmiller222 and others.
My assertion is that people are Upvoting on Steemit for the wrong reasons. They are betting rather than sincerely voting.
I don't blame people for voting this way either. The incentive is there to vote for what you think will be a hit, rather than for content which genuinely appeals to you. It's hardly ideal though, nor is it sustainable IMO. This warped incentive is why we have so many posts about "why do people Upvote shit"? etc. It's not, in my opinion because people have terrible taste, it's because they are sport-voting, betting, gambling.
Here's what I wrote:
People are voting on "sure shots" for the sake of curation rewards.
It's like going to bet on a horse race, but the favourite, who has won the last 5 races, has the same long odds as the outsiders. Naturally, you bet on the favourite.
Something needs to be tweaked to encourage people to actually vote for what they like, rather than voting based on the "form" of the author.
As far as I'm concerned, this is one of the most pressing issues on the platform right now because the current paradigm is so extremely polarising. It causes a feedback loop at the top of the foodchain.
@smooth contested
No, people are voting on "sure shots" because they don't understand the curation rewards.
He's absolutely right. I agree, for the most part, people must not realise how little reward they will get when voting late on a trending post. They still are, however, voting for the wrong reasons, and the feedback loop still occurs.
As far as solutions go, I believe Steemit needs to incentivise more sincere voting, and perhaps increase visibility of posts by "unproven" authors.
I would not be in favour of a "2 types of vote" system which some have suggested. I feel it could add needless complexity to to the users.
I'll refer to the ideal type of sincere voting as "genuine content mining" from here in. I think the system should be tweaked in the background to do one of the following:
- Incentivise genuine content mining by giving a higher ratio of curation rewards when the author has made less than X in their X most recent posts. Think of it this way; an unsigned rock band would be happy to give their agent/manager a bigger cut, for the opportunity to get their name out there.
- Incentivise genuine content mining by giving people a quota votes for under, and over, a certain threshold. For example, when you've run out of Upvotes for, say, REP 60+ authors, you can still vote, say, 20 times for authors with REP <59.
- Incentivise genuine content mining by removing the Upvote button from the listings pages. People should have to at least open the article before Upvoting it.
These are just ideas. I'm not a developer or math guy! I'm interested to know other peoples solutions, but I do know that something needs to be done if we sincerely want the best content to reap the most rewards.
I also think perhaps people should have a small allowance of flags/downvotes per day that can be cast without affecting their Voting Power. This would encourage a certain amount of dissent, which I believe it sorely lacking on Steemit at the moment.
And on the subject of visibility, I like the idea of random posts getting more visibility, but users still deserve access to the genuine "hot", and "trending" posts. Perhaps there should be a "wildcard" or "recommended" feature on the site somewhere.
I thin removing the upvote and value amount from the feed is a good idea. They should only be visible after you click the post.
Agreed 100%
You can remove it from the Steemit front-end however that only advantages those that know how to get the information directly from the blockchain itself. There are already sites where you can get better information about who is voting and how than you can on Steemit (e.g. https://steemstats.com or https://steemd.com)
Yeah very true. And i must honestly say that i'm "betting" also for some degree.
But that's what the system provokes and rewards.
The rest is voting what i really like and what i think needs to be pushed.
But i would really like to see a solution for this problem too as this doesn' t help creating good content longterm as too much content is upvoted just because it could trend. That's just betting on trends and will demotivate people who really put lots of work into their posts but don't get recognized and others get autorecognized and upvoted. Thanks a lot for taking time to show up that problem.
If the whales bet on what they like, gaming the system will reward those who can predict what whales like. If you don't like the gaming aspect of the platform, the problem is that the whales are not diverse enough to like a wide range of post. ( I am not blaming them but they are probably better coders than curators) . If the whales vote for better content so will the people chasing curation rewards.
Funnily enough, I've only started following you because you replied to my comment and you've just posted this.
Just so others can see, I sport betted on this post here, not because I particularly liked the post (i did like it, but thats irrelevant).
https://steemit.com/introduceyourself/@proskynneo/finally-coming-out-of-the-shadows-an-introduction-from-a-dev
I partially agree, however, this is the problem with financial incentives. People will naturally chase the incentives, often not to become rich, sometimes just for the sport. Catching that post at 24mins was like catching a rare pokemon.
Luckily people can vote on things they actually like as well as predicting what will be successful. I also think that while we have this situation with a few accounts holding most of the voting power, it's easy to predict what will go 'viral'. Steemit will outgrow this as the power evens out over time (I hope)
Either way, this is a very interesting topic, and I'm looking forward to reading more from you...........
By the way I will upvote around 27 Minute mark, I've got a feeling this one will get into the low hundreds. #steembetting #upvotes #curation
You've been right in the 1 time I've seen you make predictions so far. So you have a 100% success rate. What could possibly go wrong?
exactly, sometimes it's just a bit of fun. I have seen some Amazing articles on here, but the articles I've really enjoyed are usually very low earners. Steemit has already built its own 'popular' culture with 'whale' chasing, celebrities/playmates showing up and earning big. That's where the money will go. The current curation setup only aggrevates that.
Many people are simply scanning the hot queue for articles below the 30 minute threshold and betting on the likelihood that some of them may trend. I'm sure the developers are aware of this and hopefully working on a solution to mitigate it.
I'm giving away another $50 Steam Card, come play:
https://steemit.com/steemit/@cryptobarry/i-am-giving-away-another-usd50-cdn-steam-card-on-steemit-or-pick-a-number-from-1-1000
Agreed. I've done it myself. No reason not to. If I see a post by @dan that is 15 minutes old with only 8 votes, I'll hit it and move on. It's a pity that such behaviour is incentivised at the moment.
Posts from dan/dantheman will always earn very little curation reward for anyone for the reason I explained in the comments on his post. A large portion of the curation reward will go to him due to his immediate default self-vote and most of the rest will still go to him because of people voting <30 minutes.
These votes are probably a complete waste of your voting power and you would do better simply picking some unknown with good potential and taking a chance on it.
Let's look at the curation reward details for his latest post, which you voted at minute 7 (not terrible timing; a whale voted 4 minutes later):
https://steemd.com/steem/@dantheman/people-rank-using-page-rank-algorithm-for-better-curation-and-rewards
Your weight is....0.00%
What a lot of people who haven't been around since the beginning don't realize is that most of the issues that are being brought up now were already brought up shortly after the launch and solutions implemented. It's still not perfect of course, but it really isn't bad.
We try not to upvote dan the man posts as we know it will hit by too many people since we noticed as we been here maybe 2 weeks so far. Saving the upvotes for those that brings value to community now or later.
I obsessively read his posts, but I have to admit that I often upvote before I've read it...
Yup, I'm guilty too and it's hurting the platform.
.
People are use to "liking", but don't seem to understand the importance of "voting" yet. The two sound similar, but voting has a greater importance.
To give a clear example, curating comments is often handled wrong. I explained it in my recent post.
I often see someone say "Hey! I love your comment, you said it really well and I fully support your view, etc"... yet no upvote. Why? Simply because it's a new system and they don't have that new habit yet.
I only upvote what I like it think brings value. I genuinely believe I've much more to gain via pro-social usage than extracting little bits of extra curation reward.
I'm going that way too. Once you realise how small the curation rewards are, you start being more sincere in your Upvoting practices. Well that's how it is for me anyway. And that harks back to @smooth's initial point that people simply don't understand the curation system.
Blockchain solutions may be preferable but I think a UI solution could help here. The format of the homepage could affect what is seen greatly. Instead of the latest top earners, show active or trending posts from a limited amount of the most popular tags. For example, every 15 minutes, randomly choose a few of the most popular tags out of the top 20 tags and show a few of the trending posts from those tags. Show the top trenders below all of that. This way people are exposed to more content but very popular topics can still easily be found. This should help spread out the votes to more content since "picking a massive winner" will make less sense if a wider range of content is being upvoted. And since whats featured on the homepage would be somewhat random, people wouldnt be able to predict what category to vote up so would presumably be more honest in voting on what they like.
That sounds good. And couple that with removing Upvote buttons from listings. A step in the right direction, at least.
This essentially exists. The penalty for voting within minutes 0-30 along with the advantage for voting first means competition to vote first on posts which are obvious winners will pay very little curation reward.
To pay a high curation reward the post has to get to nearly the end of the 30 minutes without any heavy voting. That only happens when the author and/or subject of the post make it unclear whether it will earn a high reward or not.
In the end, the best rewards go to people who find high payout posts which were not obvious to everyone else. This will be an even stronger effect once more people understand the system and the competition gets tougher i.e. what you call genuine content mining.
Hey @smooth
Glad you found this post, thanks for taking the time to respond. That's a great point about the 30 minute thing, and indeed, it's a very elegant solution. Unfortunately though it hasn't been effective enough just yet from what I can see. People are still incentivised to camp out on "sure shots" and choose their moment to strike, regardless of the content, again betting, not appreciating. Do you think the current system is adequate? If not, I'd be really interested to hear your suggestions.
I think the 30 minute penalty will become more and more effective as the system becomes more competitive. You will only be able to get significant curation rewards by finding non-obvious good content. Dan's posts are already almost entirely worthless from a curation-rewards perspective and as both human and bot voters get smarter, more and more of the "sure shots" will fall into that same category.
I didn't realise the penalty goes to the author. That's kinda cool.
Nice, thanks for clearing that up. What does happen to the penalized amount of the curation award @smooth ? Does it go to the parent author or is it simply not paid out ? Let's hope this system really becomes more effective, at the moment it seems its worth to take a penalty for some posts because you still get better curation rewards than with voting on posts which most likely will not be discovered by someone with a little bit more power. Btw...Does the curation reward grow after that 30minutes ? I mean would you get a bigger reward for finding a post after 10h which wasn't trending but suddenly starts trending after my vote than when i vote for it after 30mins directly ?
The penalty goes to the author. This rewards authors who are getting votes based on their reputation. The curation reward doesn't grow after 30 minutes; from that point forward, the full reward is available.
I was wondering where that rank article was at for a couple days and what it meant. Thank you for sharing that in your post. You may not be a developer but you have a vision or feedback that is just as valuable which is the reason your on our feed. This is coming from a developer's point of view.