Proposal to Remove Curation Rewards
I would like to submit a proposal for the communities consideration. This proposal would not impact any existing or pending rewards and if implemented would only impact future use. In other words, this proposal will grandfather in all pending curation rewards.
Background
The reason for curation rewards was to incentivise link building. We want to reward people who sift through new stuff and discover it first. These people provide tremendous value to the platform. The challenge we have faced is that rewarding this behavior simultaneously rewards undesirable behavior.
The undesirable behavior is "bandwagon voting". This is a situation where people write bots that will automatically jump on the bandwagon of other voters. These bots can be as simple as following popular authors, or they can be as advanced as using the "recommendation algorithms" such as https://steemit.com/@dantheman/recommended to predict what whales will vote on based upon the early voters.
These kinds of algorithms don't have to be perfect in order to generate a high profit for low investment. The side effect of this behavior is a misallocation of resources (away from content creators and legitimate voters).
Potential Fixes
One way to potentially fix this system is to disable curation rewards for the first 4 hours after a post is made. This will allow the natural curation to take place at no cost to the network or the authors. After 4 hours it is unlikely that a post will suddenly become popular. Rewarding curation at this point may be a way to find stuff that fell through the cracks.
Even with this addition we have the moral hazard of people voting to collect rewards from their own curation.
Conclusion
We know that unincentivised curation works on all platforms. We know that Steem already has additional incentive to curate derived from the joy of paying the author with a vote. Any gaming of the curation reward system will detract from the platform.
IMHO this is only an issue for the early days because there are only few prominent individuals around. Once people have to pick from a larger set of prominent authors, then things will even out more. Remember, you can only upvote up to 20 comments in 24h with your full Steem power. If people run bots and end up voting for more posts than they won't make as much and can only jump on some of the bandwagons.
You are correct. Even now, it's possible to comb through posts and find some gems. As this grows, there will only be more opportunities for average people to make an income from curating content. DO NOT remove curation awards.
I agree curation rewards should not removed but TWEAKED if possible !
How does one become "prominent" in steem?
How did VIP's in the real world become VIPs?
The deliver extraordinary service!
My proposal is to equal-weight votes within the first 60 seconds or so (or even maybe 5 minutes, but not so long as to remove the value of being early). 4 hours is much too long in Internet-time (news cycles and attention span). Currently there is a large incentive to vote first, which will incentivize fast-voting by bots (even more so than now -- most of the early-adopter bots self-impose a delay). Even your own post may be upvoted by a bot before you are able to do so! By limiting the early-voting incentive to a human timescale there will still be bots but they won't have a major advantage over humans who are actively curating posts.
I'd also suggest effort into designing an incentive for downvoting. The deterrent to swarm-upvoting that is disconnected from the merit of the post should be fundamentally the threat of being downvoted later by smarter voters, which should then be rewarded (perhaps half of the reward that would have gone to a votes of previous-upvoted post instead goes to the later consensus downvoters). The possibility of being downvoted and burning vote power already exists, but it relies almost entirely on altruism, which works to a point, but likely doesn't scale very well.
This kind of knee-jerk reaction to a perceived problem reminds me of the troubled days of Bitshares, where things like "The Merger" happened very quickly and turned out to be terrible ideas..
On topic: I was thinking more or less the same thing: reduce the "first-vote" advantage by allowing a grace period like you say, the exact length can be tweaked later.
I'd also like to add that I think removing curation rewards is a very bad idea, as it removes the major incentive for most people to come here and vote on posts. Everyone can't be content creators, but if they're the only ones being rewarded we will see a quick concentration of Steem Power in the hands of the top posters. Come to think of it, why should content creators be rewarded with 50% Steem Power? Creating content does not make them good curators, so to me it would make more sense to have a different distribution: 25% Steem Power and 75% Steem Dollar for posts, and 75% Steem Power and 25% Steem Dollars for curation for example.
Before removing curation rewards completely I would highly recommend tweaking the current system, perhaps using the idea above, combined with a tweaking of reward distributions between posters and curators, 25% to curators and 75% to posts for example instead of 50/50.
Agree about knee-jerk reactions (as I said elsewhere there is hardly even a userbase yet, certainly not a representative one). Interesting point about allocating the rewards differently between SD and SP.
If equal-weight votes in first x minutes/counts, the strategy of bots would be just following the whales' or masses' votes before the period ends.
There is far less gain to that than trying to vote first. At best you get a small share along with many other people/bots. May not even be worth the consumed vote power.
I can't see how any system that differentiates between votes a very short time apart does anything other than heavily favor bots over humans. This is independent of other voting factors.
A small share? Aren't bots mostly used by whales?
//Edit: I see no issue on bots used by minnows.
Holdings have a return on investment in this system. That's fundamental in how it is designed, and will apply to any system whatsoever that avoids sybil attacks by weighing rewards by stake. The rich get richer is not, by itself, a problem, as long as it doesn't negatively affect categorization of posts, or appeal to a wider user base.
By "small share" I'm comparing it with the enormous share that the very first votes get in the current system relative to the later votes. That strongly incentivizes bots.
In actual practice all the current bots run by whales delay their votes 30-120 seconds and are rarely first, in fact often quite late in the mix of votes. As hard as it may be to believe, current whales are not trying to wreck the system for their short term gain. This by the way does not include me at all; I don't run any bots.
WHAT?! You have got to be kidding. This is June first, not April first... This will kill it.
If anything, rewards for normal users need to be increased enough to make it interesting for EVERYONE.
Every post and comment and vote should give a user something, and users should be able to tip what they like too.
Otherwise this is just another board...
Well said. The pitch for this site to users is "post, vote, get paid". Many users of any site are lurkers. They won't post much, even links, but they will vote and many will sign up based on the appeal of getting paid for voting well (also the lottery aspect of voting early for a post that becomes incredibly popular e.g. front page on reddit).
Removing the curation rewards will kill 90% (or maybe 99%) of the appeal in terms of user signups and differentiation in a crowded market.
I like that idea ...
If I could choose how much of my voting power/Steem Power/STEEM/SBD to give to a content creator I would have a little more control or options from which to choose. Say I see a really good post and I can click a dropdown from 1%-100% (that adjust as I vote during the day indicating how much "power" I have left) and assign a voting power amount that I deem worthy of this post.
For instance, if I want to be a complete tool and give my post a 100% upvote, well I have no voting power left for anyone else and so be it. Or if I want to give 10 people 10% of my voting power then that's what I choose from the dropdown.
And perhaps also have the option for me to tip as well, give them a little something from my own personal stash of STEEM or SBD for the good content.
HI
Removing curation rewards, or detatching them from the user's Steem Power, would reduce the incentive to vest your STEEM, wouldn't it?
I upvoted this comment but I will also answer in the contrary. Presumably the weight of your vote would still be based on your SP so there is an incentive to buy or own SP in terms of being able to improve the visibility of content. But the only people with a real incentive to do this are effectively advertisers, so it becomes a bid-for-placement platform, something which is already a part of the value of SP, but which coexists with the potential for return-in-investment via curation rewards.
I expect this to be destructive to the value of Steem both in terms of the capitalization and also the usability and appeal to readers
LOL - I'm sorry, but my vote is really small :-D. I have voted a hundred times today.
That is why we need separate voting power for comments on each post !!!
That was quick.
Removing half of the rewards the platform is based on at the first sign of trouble does not inspire confidence.
Preventing the bots that are causing the problem, and putting more effort into balancing the the effect of regular users vs whales for curation rewards would be more useful. Multiple regular user votes should be balanced to have the same effect as a whale vote, or the regular user will stop voting. Every vote should count for something especially en masse, not just the insiders.
If you want to fix this without scaring off the userbase, a public request for proposals and a vote on the popular ones would inspire more hope. (more than, hey this isn't working out I think I'm gonna scrap half the reward system).
People have invested time and money based on the rewards promises, how you handle this will influence future investment.
Proposal coming...
The more invested monetarily in steem the more you have to lose. The more you have to lose the bigger you influence should be. Your vote in a company you only own 1% can't worth as much as someone who own 98%. A competitor would just by 1% worth of the company and vote stuff to make the company go bankrupt. The weight of your vote must be proportional to what you have invested monetarily or otherwise.
There is definitely something wrong with curation rewards. I don't think it is the bots, though. I think most of the voters (including me) just quickly skim through the article and quickly upvote if it seems promising. You can always change your mind later, but this rarely happens. This strategy is quite natural, as the time plays so big a role.
I still think curation rewards are a necessary part of the network, as only some people are content producers, but the vast majority are consumers / voters. We might look for a better algorithm which will incentivize actual reading of the articles and informed voting. If we can do this and attract more content creators, bots will not be a problem.
Removing curation rewards will remove our consensus mechanism. Politics and collusion will replace it. "Vote for me and I wont downvote you"
The reward algo together with the auto up-voting phenomenom create an effect that is closer to
votes^3thanvotes^2. Could the exponent of curation rewards (and maybe even content rewards) be reduced tovotes^1.4? Would it enable sybil attacks?Maybe a random "timer" (e.g from 0 to 8 hours) for every new thread/topic that is created?
If for example the random "timer" for your OP was at 3 hours and 21 minutes, insta upvoting from bots becomes useless. The lucky upvoters that upvoted closest to 3h21m would earn the most in this case and not the bots. Bots would loose their speed advandage and now we know that we have plenty of time to read and evaluate the content. Long term we earn the average and there is no time pressure.
I think the developers should consider this idea very seriously! I think it is brilliant idea!
Clever bots will know when the timer starts, then vote afterwards. Unless .. calculation of rewards happens at an even later time point and it decides which votes are qualified for rewarding. But uncertainty will probably harm adoption.
Maybe first round calculations can be done at the end of the 8 hours timeline, also if the "random timer" can't be kept secret until then, it could be created at that time too. It adds some randomnes but I think that rewards can continue normally to be paid and bots become completly(?) useless.
Also, voting one hour earlier or one hour later from the "random timer" should have the same "value weight"/reward
How will they know? The timer is "random"!
Why uncertainty?
!!!!!!! IDEA !!!!!!!
Let the votes before the random time elapse.... COUNT!.... BUT ONLY 10% of their original VOTING POWER! (could be any % the community decides)
On a blockchain, even a random value should meet consensus, which means every node (witness or miner) would know the exact (SAME) time, of course the bots would know it as well, and take advantages.
Read @abit's reply and then basically remove the word random from your thought process when considering blockchain systems. That's isn't 100% accurate but it is close enough to stop making major errors in reasoning about how they work.
but can the bots not take back their votes and vote again until "they know" they voted after the random time elapsed? I guess the random time must be revealed only after payment...
I think curation rewards have to stay. It's a kind of labor and it should be rewarded. Sure at some point some algorithm can use conditional preference networks to generate recommendations but that takes a long time to learn what to recommend and often it's not perfect. A human based evolutionary computation is what we have now and curation deserves reward.
I prefer we keep curation rewards whether human or bot collect the reward. It may be possible to improve voting but a less than perfect mechanism design is still better than no mechanism design. Curation rewards for a slightly flawed algorithm is still better than no curation rewards. Finally we probably should wait until the network is bigger so there is a larger sample to work with before determining to make any decision as extreme as this.
Some thoughts,
Do we know the public sentiment? A few people have decided that the curation system with bandwagon voting is having a negative impact on the functioning of Steemit but how does the negative impact express itself in a way which can be measured? Are we likely to see less growth in content producers in the short or long term if bandwagon voting remains? Do bots really have a negative effect in the long term if the population is big enough so that there are a diversity of bots and whales? It could be that under the current conditions the mechanism design for curation voting is not good for the current demographics but what happens when there are 100,000 users with bot swarms and whales?
Personally I think if the bots swarm and there are whales but it's diverse enough that it could offer a level of predictability so it's not completely random but also be not so predictable that people everywhere can game the patterns. In essence there could be many different patterns for bots which can be encouraged. In generally I just don't have any clear information as to the current effects or predictive statistics so I can't make heavy decision. I request more information on usage data or at least some kind of possible projections, a hypothesis, etc.
Perhaps a better compromise is to reduce the allocation to curation to until the potential for abuse is better understood.
Reducing curation rewards without changing the algorithm will not change the behaviour of voters. There will still be the bandwagon voting behaviour.
Agree. It will also further favor bots.
Right now the highest Weighting goes to the first Upvoter and diminishes from there. If possible place a random element in the code that randomly shuffles the weighting allocation so that the the element of uncertainty is introduced for all votes received by the post. Then even a late post may earn a high weight and a early vote a low weight. Therefore voting rewards becomes less predictable and curators are less able to "game the system"
OOP' s I see liondani proposed a similar thing in the later comments
That would incentivize to spend your votes on the 20 most valued posts
Couldn't the people with 40+ accounts have all 40+ accounts upvote post and game the system that way?
Anything that is weighted or chosen by "accounts" rather Steem Power is broken IMO. You could choose randomly but weighted by SP (so 2x SP would give 2x the chance of being chosen), although doing anything "randomly" on a blockchain is problematic.
My proposal is to give equal weight to all votes for some reasonable period of time (one minute, five minutes, whatever) which makes it possible to at least read the post before voting on it without being penalized for doing so. By rewarding specifically the first even if that occurs in seconds you strongly favor bots. Later votes would continue to have declining weight to reward those who first recognize the value of a post as well as (though by less) those who later endorse the earlier votes.
@liondani. There is nothing to prevent voting after the 1-5 minute period, even 12-24 hours later, and it should still be rewarded. It just should no longer receive the "early voter" bonus that is intended to reward those who find new unrecognized posts. Once a post has already started to receive votes, it is no longer unrecognized. My proposal is just to stop rewarding the very first post (or a few of the first) by enormously more than those who vote just a few second later. There is very little value there.
A reasonable time ok BUT RANDOM.... else the bots will vote exactly after 1, 5 minutes or 10.... they will not care if a minority of users voted before them....
And I don't agree 1 and 5 minutes is reasonable in a 24 hour period... I find more reasonable 1-2 hours before the original voting power kicks-in. We want average users to have the opportunity to get better rewarded ... And average users are not logged in 24 hours a day to take advantage of 1 and 5 minutes windows... I mean average users will have a bigger selection and will vote with less pressure if they have more time to read the content before voting....
Look what happens now, we upvote most of posts because we read a nice title (because we are in hurry off course) and then we read the underlying content :)
That sounds better. If you strip away curation rewards, what is the incentive to spend time on the site? How will topics be curated? Whales will just sit on their stashes and not use them to better the project. People will get lazy also. It will just be like Reddit, etc. Some compromise would be far better.
That could work. I request more study and more data so we can know what the effects are.
deleted
Astute point about not really knowing the effect. Most of the bot voting we've seen is for authors who actually make good posts. That generally speaking adds value and is not necessarily a negative. The fact that someone made a post that was deliberately designed to game the bots does not mean there is an overall problem.