Design the rules: Inside a linear rewards distribution mechanism
It seems that many people didn't understand how a linear reward distribution will be working, they just think it won't work.
Here I'll write how it looks like.
In short
Stake weighted voting. "Number of votes" below means quantity of SP times voting weight.
For authors, reward is number of upvotes received minus number of downvotes received.
For curators, reward is number of upvotes contributed minus those were counteracted (downvoted).
Effects
Each SP has same weight, no matter voted on a popular content or not, no matter when voted, no matter voted by a whale in a mass or voted by a minnow.
Up-voters will vote on contents they like or think is good or worth voting, no matter when, no matter what content, earn almost same rewards. The only exception is, when others downvote a content you upvoted, you earn less. So people is encouraged to vote on better contents.
The ones who don't vote earn nothing. So engagement is encouraged.
Downvoters earn nothing. They are volunteers.
Bots will be trained to upvote for contents with higher upvote/downvote ratio, probably right before payout. So they'll cause the least (negative) impact to the trending page.
Volunteers will check before payout and downvote overvalued contents accordingly, to fight abusing.
The key to succeed
Quality of downvoters who give up curation rewards.
They are justice then the system would be justice.
Examples
Alice wrote a post, a whale voted with 3000 SP, then Alice will earn 3000 SP and the whale will earn 3000 SP. The post's downvotes ratio is 0%.
Bob wrote a post, 100 minnows voted with 1 SP each, 2 dolphins voted with 100 SP each, then Bob will earn 300 SP, the minnows will earn 1 SP each, the dolphins will earn 100 SP each. The post's downvotes ratio is 0%
Cid wrote a post, 10 dolphins voted with 100 SP each, a whale voted with 1000 SP, another whale downvoted with 1000 SP, then Cid will earn 1000 SP, the dolphins will earn 50 SP each, the whale upvoted will earn 500 SP, the whale downvoted will earn 0. The post's downvotes ratio is 50%.
Dick wrote a post, self-upvoted with 10 SP, some people don't like the self-upvote so downvoted with 10 SP, they will all earn nothing. The post's downvotes ratio is 100%.
Emily thinks Dick's post still has value, so upvoted with 10 SP, then Dick will earn 10 SP as author and 5 SP as voter, Emily will earn 5 SP as voter. Dick's post's downvotes ratio reduced to 50%.
Fox is another whale who thinks Alice's post don't worth 3000 SP, so he downvoted with 1000 SP, then Alice will earn 2000 SP and the earlier whale will earn 2000 SP. Fox himself will earn 0. Alice's post's downvotes ratio is now 33%.
Gary is a bot with 50 SP now thinks it's time to vote, he'll vote on Bob's post which has 300 SP only but nobody downvoted, so it will give Bob 50 SP more reward and earn 50 SP by himself as curation reward. Bob's post's downvotes ratio is still 0%.
Harry is a dolphin with 150 SP who voting for good contents, he thinks Bob's post is undervalued, so voted which will give Bob extra 150 SP and himself 150 SP. Bob's post's downvotes ratio is still 0%.
Ivan is another mini whale with 300 SP who voting for rewards, so he voted on Bob's post which has no downvote, which will give Bob extra 300 SP and himself 300SP. Bob's post's downvotes ratio is still 0%.
Jack is another mini whale with 200 SP who voting for contents and thinks Bob's post is overvalued so he downvoted, now Bob will earn 600 SP, total upvotes is 800 SP and total downvotes is 200 SP, so downvotes ratio is now 25%. So the minnows voted with 1 SP will earn 0.75 SP each, the first dolphins voted with 100 SP will earn 75 SP each, Gary bot voted with 50SP will earn 37.5 SP, Harry voted with 150 SP will earn 112.5 SP, Ivan voted with 300 SP will earn 225 SP, Jack will earn 0. Bob's post's downvotes ratio is now 25%.
Ken is a dolphin with 30 SP thinks Dick's post has the smallest pending payout so less likely be downvoted again, so voted on it, then it will give Dick 30 SP more as author reward, so in total 50 SP upvoted and 10 SP downvoted, so downvote ratio become 20%, so Ken will earn 30 * 40/50 = 24 SP, at the same time Dick will earn 3 SP more (change from 5 SP to 8 SP) as voter and Emily will earn 3 SP more as voter as well. Dick's post's downvotes ratio is now 20%.
...
Some people think all posts are good so vote for all.
Some people don't read posts but vote for all, or randomly.
Some people don't know which post will have the highest reward so vote randomly, or the ones they like the most.
Some people vote for the first posts they've seen because no time to read others.
Some people didn't vote on the first posts they've seen because no time to read others.
Some people think some posts are overvalued so downvoted.
Some people think some posts are overvalued but don't want to waste voting power so voted for others.
Some people like some authors so have bots voting for them only.
Some people trust some voters so have bots following their voting.
...
Near the end, before payout, some bots voted for posts with lowest downvotes ratio.
Then in the review period, volunteers identified some votes cast at last minute were abusing so downvoted accordingly.
Some smart bots know they'll be downvoted if voting at the last minute, so will vote earlier but will be less sure if they'll earn the most rewards, although the post has a low downvotes ratio when they're voting.
Other smart bots know they'll be downvoted no matter when, so will vote for most popular contents to minimize the impact of downvotes.
The end.
Miscs
Hope someone can explain the mechanism better than I did. If you know one has been done, please tell me the link.
If you think the mechanism won't work, please tell me why.
Thanks.
Earlier post: https://steemit.com/curation/@abit/benefits-of-pure-linear-reward-distribution
Excellent :)
Maybe even better to show the somewhat complex scenario (Jack and Ken) in a flow chart, or spreadsheet. I'm a formula person, and like spreadsheets and charts since they come easy to me; others may rather like the way you formulate and present.
I must say, I tried to do a spreadsheet quickly, but failed, too many columns. Therefore maybe better in flowchart? When I can find some time, I can try to do that tonight. Or you have a go with it? :)
I have no go.. I'll appreciate if you have it done. :)
For those who likes Use Cases described by @abit in post above in table format, they can be found below for the 2 extended use cases:
Note: each example given by @abit for both use cases Bob and Dick are given as a separate scenario in separate columns in the tables.
Use Case: Bob
![image of use case Bob](https://steemitimages.com/640x0/https://i.imgsafe.org/dbb738b7cb.jpg)
Use Case: Dick
![image of use case Dick](https://steemitimages.com/640x0/https://i.imgsafe.org/dbb751d734.jpg)
Thanks for that.
Hope it'll benefit others; It did for me :)
@abit resteemed! luckily I bumped into this at @oaldamster's page
Thanks for explaining it so .. almost idiot proof!
It helps specially if one is a newbie!
When I was a newbie I didn't know what to do .. it was like swimming in a school of a variety if fish and I thought just vote whichever was in trending .. but later on I have managed to learn swimming in the current and finally decided to just upvote the articles I really like and sometimes though I don't like the article just because the author is very interactive I upvote that authors's post, too!
We'll never know whether it'll work till we allow it to be executed for a certain period of time and I mean at least more than half a year.
Thanks for the reply.
I'm not a native English speaker, so had difficulties to explain it well.
In my last sentense, the first "working" refers to the internal mechanism / algorithm, the second "work" means "running well".
On the surface, this looks fairly clear and simple and understandable... "on paper," the mechanism would probably work.
Where I have a question mark is in terms of real world building and long term sustainability of a COMMUNITY. This brings into question the relative value of a bot vote vs. a human vote, and the psychological impact that has in terms of both attracting new users and retention of those (and Older) users.
Over and over, I see posts with 250 upvotes and maybe 25-30 views. Stated differently, I am now looking at content that has 250 automated pings, and 30 real human interactions.
Don't misunderstand-- I have nothing against bots. They have a definite purpose, whether it's @cheetah or the twitterbot, or bots designed to discover spam, cheating, abuse, keyword stacking or whatever. Even bots designed to "fetch" promising content for future interaction, or "welcome bots" for new accounts (only). The problem arises when we decide that a random bot upvote has the same "value" to Steemit as a real human manually reading and curating the content, and perhaps leaving a relevant comment.
Bots are-- in essence (unless you're a very advanced developer)-- "stupid." At their root, they determine simply "content exists." They don't "have opinions," they don't "like or dislike," they don't "have insights;" they are not "moved by;" they don't "learn something" from content.
Again, not saying bot should be abolished... but asking whether it is REALISTIC-- in terms of new user attraction and retention and building a thriving community-- to have a system in which a bot vote has the same relative weight as a human interaction?
Maybe this comes back to the discussion of Steemit's self-identity. This is allegedly a social content site. That's great but... if the platform is dominated by an "army" of non-human automation, doesn't that directly contradict the whole notion of "social" and "community?" The counter argument is *"Well, a lot of people don't have TIME to interact with all that content!" FACT! But this is a "social" site, not "Game of Bots."
Let's translate this to a more human perspective. You live in the town of Steemit which has 10,000 people in it. You're one of the city founders so you're a "Whale." You don't have TIME to go out and shake everyone's hand everyday and congratulate them on their new flower garden. So you you send everyone a postcard that says "Hi, I remember you!" (That's a bot). Should that really carry the same weight as a real citizen who goes by someone's house and spends 15 minutes looking at the flowers, chatting and sharing a cup of tea?
My point isn't that the "postcard" approach should be invalid, but merely downvalued in order to stimulate growth in the now stagnant new posting rate, and to stimulate actual interaction on posts... which will encourage people to create more "actionable" content... which will look attractive to a potential new user ("This is like Facebook, only cooler!"), in turn looking attractive to investors, in turn helping bring the price of Steem back up.
I'm a content creator, not a developer. Part of me is very excited about the potential of Steemit as something truly revolutionary... on the other hand (even after just 30 days) part of me feels a certain frustration that technology here so dominates the human factor...
I do not like bots voting for content as well. Only when bots get that intelligent that it can behave like a human being and therefore identify 'quality' content. But I suppose bots are not able to do that until AI/humans reach singularity. But that takes still a little bit of time, to some decades, to others a century, to a small group never, only time will tell :)
Not sure if self-identity or identity management will solve the issue to determine if an account is using a bot for voting, or used for manual voting, or a hybrid. In the end a bot could filter messages towards the bot manager and Steemit account holder, after which the account holder can still manual vote, or let the bot look like it is manually voting; For instance the bot will randomly in time vote for content with some rules around it so it still tries to earn sufficient amount of rewards.
A way to remove bots from the network is to remove the curation rewards entirely. Then absolute no incentive is available for bot voting. People than say that there is also no incentive for individuals to vote, but I personally believe that individuals will still vote. But hey, that is theoretical, we shall see it in practise. The wanted bots (see above) can be paid differently, maybe a small percentage of the total minted Steem can be used to pay for the bots for development and maintenance of these bots, or all bots shall be created by Steem INC and paid by the Steem INC account.
Regarding identity management, I'm in favour of open real identity, so no real identity hiding. This would then also allow to set a maximum to the voting power, ie whales can be made less powerful by design.
But I do understand the resistance against open real identity. So maybe implement a solution for that? Although I'm not a software architect or software developer, I think a solution could be to route vote request to a 3rd party entity who does the ID mapping; that entity should have mappings of real identity to all the user IDs (no idea what this will do to performance such as response times but I guess this can be very quick with sufficient computing power, database speed, and internet bandwidth). The data shall be encrypted in the database, so nobody can have a look at it. That entity will not be open (ie not decentralised and information not on a blockchain), therefore such entity shall be somehow 100% trusted. Such entity could be run by people that get elected by Steemit community and every so often re-elections are held. Maybe some 100% technical solution is possible, but I'm by far not the right person to be able to design a workable solution. But my gut feel says that somehow a solution can be created for identity management. Maybe there are more technical individuals reading this message who can figure out if a solution can be created, and what the consequences.
Thanks for making it as basic as possible. As a several day user, and generally not too clever, I am getting a handle on the way it is, and the way it will change. Einstein said if you cant explain something simply, you don't understand it well enough. I think you have done well. :)
It is actually easy to explain, as it obfuscates nothing; won't that be considered a negative?
I'm still two cups of coffee away from shedding my morning cynicism 8-).
Of course it will work, the only side effect I would expect is that top-1 post having now 200+$ payout with new rules would get something like 50$ only.
Almost think you are going to go down to Tom with Dick and Harry being mentioned.
This depends quite a bit on downvote and I think if there's any reason it won't work will be it's not in our habit to downvote. At least not now.
Why to Tom? There is no name begin with U?
Actually I'm not sure whether there will be enough downvoters and whether they'll all be good / justice.
You can go to Z but to Tom means you get Tom, Dick and Harry
https://en.wikipedia.org/wiki/Tom,_Dick_and_Harry
Oh I didn't know the Idiom. Thanks!
You're welcome
Very simple and perfect rules, thank you for bring it to us.
As in the case of bots, I don't think it's a problem at all.
Although contents with higher upvote/downvote ratio don't always mean quality content, but most of the time they are, so they deserve more.
The next thing we'll discover should linear reward distribution be implemented is that people still complain about the distribution and wealth disparity between users.
Comments complaining about the number of votes vs views will continue to draw discussion. But really, the whole issue boils down to people mis-interpreting the buzzwords that the whole crypto community love to throw around - Decentralisation.
Any kind of stake weighted system of votes, linear or polynomial will only skew the curve of the existing imbalance of wealth distribution. (Which itself is natural and I have no issue with).
Should we decide to get rid of stake weighted voting completely, we might end up with a system where votes themselves are fungible, but of course we open ourselves up to sybil attacks.
This post has been ranked within the top 50 most undervalued posts in the second half of Mar 06. We estimate that this post is undervalued by $6.09 as compared to a scenario in which every voter had an equal say.
See the full rankings and details in The Daily Tribune: Mar 06 - Part II. You can also read about some of our methodology, data analysis and technical details in our initial post.
If you are the author and would prefer not to receive these comments, simply reply "Stop" to this comment.