Why n^2 ?

in non-linear •  7 years ago  (edited)

In my last post I tried to show, why I want 'non-linear' rewards back.
It seems that this idea is not well received.
Most people argue, that their votes would be worth less then.

But that's some serious nonsense.

The amount of r_shares you can allocate towards a post depends on your Steem Power, your Voting Power and the Vote Weight.
Given all other factors are the same, an account with 2x as much Steem Power can allocate 2x as many r_shares.

And this is the same with n^2 or any other 'curve.'

I am arguing to change back how the rewards get distributed based on r_shares.

With n^2 it is true, that your vote will be only worth 1/4 of the rewards of an account with 2x as much Steem Power on a blank post.
But that is a very unlikely scenario.

The whole point of n^2 is to reward stacking of votes.

If you were to use your vote on a post with already a bunch of r_shares, then upvoting that post would allow you to allocate much more rewards than if another account with 2x as much Steem Power voted on a different post with no other votes to stack on to.

The problem

The real reason, why rewards were so 'unfair' before hf 18 is this:


(taken from steemwhales.com)

A few accounts hold most of the Steem Power.
No change of the rewards distribution can change this imbalance overnight.

The trending page might look better to some now, but consider how many of those posting rewards go straight back to the whales, who sold their vote.

The Steem shares are distributed very disproportional right now.
In my view, this has been the main cause of problems and will be for quite some time.
There is just no way around it.

But we can not fix this problem by employing a broken rewards distribution algorithm.

It just does not make any sense to me.

I feel like most people prefer the change that came with hf 18, because their vote is worth more now, which you can see above is false unless you enjoy voting on comments and posts, that nobody else wants to reward. *
* abuse/self-vote

I will give you the benefit of the doubt and will assume that you are not a notorious self-voter, bid-botter or other sort of rewardpool-rapist and I will continue to blog about this issue in an explanatory manner, but I feel like most people understand it well enough, but just see their own rewards and just look at their own wallets and misunderstand these posts on purpose.

I hope to see you next time, when I post more about n^2.

Peace

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  
Loading...

I'm with you, and therefore you are my ewitness proxy. But I dont spend much time for these discussions.

I really appreciate that.

So...what exactly can we do about this, except voting for you as a witness and spreading the information?

I mean right now, it's a profitable approach to ignore the community and reward yourself by voting only on yourself (bonus ignorance points for leased SP...). In a short term view you are pretty much printing money with very low effort. People who are commited to this method probably won't just step down...because it works for now. They're also setting a questionable example for others who then might try the same approach, which could snowball out of control.

Of course in the long term view, if everybody did it like this (the german term "Heuschreckenplage" comes to my mind) steemit would fail, but everyone who did use steemit like this and already paid out his "earnings" would come ahead on top.

What can I do to support this view? Are there more witnesses who think alike? Maybe a post with all witnesses supporting your view would be an idea, because it shows that not everyone agrees on the current system. Might get a bit more recognition as well, instead of just one person posting his reasonings (of course if there are witnesses who want to keep the current system, they could also share their view and reasoning).

Thanks for taking a stance on this issue, you already have my vote though. ;)

Good idea about the witnesses who support a curve away from linear reward. We'll work on this.

The current systems is at the expense of the majority shareholders and to the advantage of everyone else. If the majority shareholders were to understand the issue they would soon vote to change the reward curve.

Hopefully this should be relatively easy to achieve if people are well brief on the subject.

I just left a comment to explain the issue.

Thank you for your support and understanding!

Thank you for chiming in to this important topic. That's the part I really like about this network. You don't feel as helpless when you see that valid concerns of the users are seen and evaluated critically. I hope that the other witnesses follow this example and explain their reasoning about the situation of the reward distibution.

In my view this is a conflict between people who view Steemit as a short term cash grab opportunity who just move on once the system fails (though I could give them the benefit of doubt and assume they are not even realizing the long-term consequences).

On the other hand you have those who believe in the value of a decentralized social network. This is a long-term strategy focusing on sustainabilty. The social aspect would fail, once even more people realized that social interaction and good content is actually not the best way to advance on Steemit. Yes, they might find friends and learn new things on here. Maybe that's enough for them, they just bury their head in the sand about the bigger picture and focus on those great parts (a pretty common strategy actually as far as I have seen).

But while they care about communication and the content of the users, those who don't do this just overtake them left and right which might in the end lead to discouragement of those users from the second group, who are the ones creating real value (and therefore stabilize or advance the overall value of steem).

Of course there are also people falling kind of in the middle between those extremes.

However, it is possible I am missing something important out of the bigger picture, so I am looking forward to the discussions. I'll keep an eye out for more opinions on this topic.

I only know of @teamsteem, who sees this problem and wrote about it.

There is a lot of work to be done convincing the right people and I think @ned needs to approve for this to ever happen and it doesn't look like he would support such a fork ...
Technically, a majority of the top 20 witnesses would need to support the new fork.

Alright. I've looked at the introduction and some posts of @teamsteem and voted for him. I hope some of the Top20 witnesses (and others as well) see this post and share their views on the situation. I thought about mentioning them in this post, but I'm not sure if they would appreciate this. I would like to get an opinion of as many as possible though, because I do agree with you and @teamsteem about the fact, that if the economics of steemit fail or give false incentives, then no amount of nice tools/inventions from the users can make up for those failings.

Steemit has a big head-start from the view-point of user base, but there are other options of blockchain-based social networks evolving. If new users see those alternatives and feel like the reward system there is more fair, they might very well decide to make the switch.

Interesting times. Thanks again for explaining your reasoning.

  ·  7 years ago (edited)

Thanks for the replies.

Some of the comments I get for this are so out of this world that it feels good to meet someone who understands the matter instantly.

17 out of 20 would be needed.

  ·  7 years ago (edited)

Edit.....another query for you.

Are there any plans to limit the amount rewarded to a comment made on an out-of-date post? or is there a reward? Have come across those who spend a lot of time posting comments on older posts.

Thank you for posting about this issue @felixxx.

Appreciated reading the article and your reply to @croctopus.

Will give teamsteem a look and read his view as well.

Wishing you all the best in your endeavour to bring this problem to awareness.

Cheers.

I just left a comment explaining my views on the subject. Thank you for your interest.

Are there any plans to limit the amount rewarded to a comment made on an out-of-date post? or is there a reward? Have come across those who spend a lot of time posting comments on older posts.

The payout window of a comment is not connected to that of its parent post.
Writing comments under older posts is fine and can be rewarded by votes.

Thank you for the support.

I completely agree that linearity is a problem, but a lot of people are clearly concerned that n^2 is too extreme. What do you think of n^1.5?

I think it's also important to bring back a serious vesting schedule so that more whales actually think about long term value when they're deciding how to use their stake. 2 years seems reasonable to me.

Now we are talking.
Sure n^x is an option.
I am mainly for non-linear rewards.

n^2 was a brilliant idea and is worth trying out again, though.

Yeah, I don't object to n^2, but the main issue is that we need an exponent greater than 1 to provide incentive for consensus. I'm also interested in curation rewards for downvotes to fight spam and curation penalties for voting against strong consensus. Interface sites should have highlighted 'controversial' sections to embrace the vote wars and catalyze consensus formation on community values/standards.

I didn't mention it in this post, but n^2 would affect flags too - flags being much more potent on overvalued material.

Rewarding consensus would also affect curation rewards indirectly.

Yeah, I get that. I still think the change to the short vesting schedule was also a mistake and that the curation system could use some additional work beyond reverting to an exponential curve.

If the only change made was going to n^2 I think we'd see a lot of whales just harvesting and no one would have the incentives to stop them. A lot of users get very discouraged by that sort of thing already.

If the only change made was going to n^2 I think we'd see a lot of whales just harvesting and no one would have the incentives to stop them.

  1. Those whales are harvesting the posting rewards either way.
    Right now it just isn't as obvious.
  2. With n^2 there would be an incentive to flag.
    If you were to try get the most impact out of your votes, flagging the most exposed material will have the most effect.

Like I wrote:

The trending page might look better to some now, but consider how many of those posting rewards go straight back to the whales, who sold their vote.
The Steem shares are distributed very disproportional right now.
In my view, this has been the main cause of problems and will be for quite some time.
There is just no way around it.
But we can not fix this problem by employing a broken rewards distribution algorithm.

All the linearity does is cloak how much money the whales can sifon out.

Hey @ned, ...

Give 'em hell, felixxx, they just ignore me,...

Thanks for bringing the subject up. So voting power varies linerarly with Steem Power now, is that it?

I left a comment giving a bit more explanation above.

Under n^2 reward curve, if you have 100 SP your vote is worth roughly 100100. If someone else as 50 SP their vote is worth 5050 or 4 times less. This is also a super strong incentive to give 100% upvote.

I gave more details about why anything away from linear reward is better that linear reward and linear reward is economically unsound.

Okay, thanks. I'll read it.

No.
Posting rewards vary linearly with r_shares now.

Okay, thanks. Do you have a link to a paper that explains the difference?

There is no paper for this, unfortunately.

I recommend you try to look into posts about r_shares - there are some good ones, but I can not recommend one.
I will keep blogging about this, too.

Thanks again. I just hit up Google for "site:steemit.com r_shares". That should give me enough to go on.

But @felixxx i don't understand why you think this will help against self votes for whales ?

If they have strong account than they will have enough r_shares. Do you think that it's a problem for them, to make few other accounts with SP delegation to those accounts to "spread" the r_shares among votes?

And bots selling votes could easily do the same. You "pay" for the vote and you get 10 votes from different accounts. Wouldn't that exploit the n^2 as well? the only thing you would do with this would be taking those couple of cents that minnows get and are more than grateful for them. Even if it's from themselves ...

But what you would get with this is, that "Steemit elders" who vote for each other would benefit exponentially as they did before. Wouldn't they?

So much talk about the reward distribution change. I believe that the community is the only one possible to handle this. Not changing reward distribution. It's the only reason we need laws and police in real life. PEOPLE WILL CHEAT if possible and it's always possible, so they need sanctions. Some strong steemit accounts are already flagging self-voters, and I strongly believe that an "authority" is the only way when dealing with greed and money. Whales who thinks the same should form an alliance similar to @steemcleaners. But that is only going to help a bit and stop those with a bit of conscience. Because those who want still can post 30 comments and upvote them last second before payout lock.

Maybe it would be smart to allow flags for the last 12 hours after the upvotes are "closed" ... That could give the "authority" some advantage for last second upvoters. Because they could downvote it in the last 12 hours.

But @felixxx i don't understand why you think this will help against self votes for whales ?

I did not explicitly mention how and why it would also stop self votes for whales, but will explain that in my next post about this topic.

Cool I am looking forward to it! :)

The downvote window is already 12 hours longer than the upvote one. Exactly those last 12 hours

So there actually shouldn't be any problems. We are given all the tools we need to stop self voting. But we as a community are not using it (yet ?).

_

I believe that the community is the only one possible to handle this. Not changing reward distribution.
Would be a first Step to stop the Plattform actively Promoting Self-Voting ...

Self-Upvote.jpg
...by removing the Prompt while Posting?

But @felixxx i don't understand why you think this will help against self votes for whales ?
Because of Whales i'm an Fan of using an Sigmoid Function:


(as statet in Detail in my Comment a Mile below from here). :D

I am not arguing for linear rewards but one of the benefits is a much more stable post payout estimate. Since Hf18 your estimated payout doesn't drop in the same way it did previously. Before I found if you got a big upvote you would get a boost to your estimated payout but it would decay away until the final payout. I found this really discouraging, maybe in part caused by the longer payout windows in the post. it seems more stable now.

good job

du würdest nicht zufällig den Beitrag nochmal auf deutsch verfassen oder?

I can go with the n^2, but as long as it isn't proportional to one's steem power. Make each vote count the same indipendent of steem power.

Your idea does make much more sense now. I think I will agree you. But if I knew more about the other side of the story then I could decide after. Still I agree that there needs to be less of self-voting (yes I know I should work on that) but voting on other peoples comments and post what we really should be doing. So yeah

Voting your own content is not the problem. ( if you think it's good content, anyways :P )
Before hardfork 18, I was upvoting my own content all the time, but my single vote was only worth 0.01 $ and only if a lot of others voted for the same post, would it generate a good payout.
Changing back to n^2, would make this page much more 'social' again.

Ohh okay. So that would make everybody's upvote worth 0.01$?

As an example:

With the rewards working like I propose above, an upvote from my account would be:
On an 'empty' post 0.01 $
On a post that already has some strong votes on it ( or when it gets those later ) the same vote would be worth 10$

Those are just random numbers, but I hope you get the idea ...

Ohhhhhh That makes a lota sense!! Thanks!!

Still trying to digest this. Perhaps a simple explanation video will help people like me. LOL

With this proposal, does it mean we should not upvote our own posts immediately after posting? Instead, upvote only after there are a bunch of other upvotes. This way, we kindof reward those upvoters before me without diminishing my vote value (compared to what is happening today)

If you were to use your vote on a post with already a bunch of r_shares, then upvoting that post would allow you to allocate much more rewards than if another account with 2x as much Steem Power voted on a different post with no other votes to stack on to.

That's genereally a nice approach if you dont self-upvote right away ;)

Loading...

I am pro linear and do not see it as a problem any more. I think linear is the best and most fair reward curve that everybody understands and also helps minnows. If a whale only self-votes and nobody else votes for him, he can only get the yearly inflation of currently 9.5% back, right? So this is not really a problem. If at all, the SP distribution is the problem, but as more and more Steemians register, whales and Steemit Inc will sell off some of their SP and distribution will get a bit more even. If reward curve is set to n^2, cheaters will just exploit that too by delegation and creating multiple accounts / multiple upvote bots. Therefore I think linear is best, HF20 will make self-votes less profitable by decreasing author reward from 100% to 75% and by reducing the 30 minutes window to 15 minutes. I think this is a fair change.

If a whale only self-votes and nobody else votes for him, he can only get the yearly inflation of currently 9.5% back, right?

What do you expect me to reply to this nonsense ?

Try to give a constructive explanation or counter argument, how a self-voting whale would get more than the yearly inflation of about 9.5% of his vested Steem if he only self-votes, and does nothing else and nobody votes for him?

A whale can make easily more than 100% roi in a year if they wrote 10 bullshit posts a day and voted on them.
Thanks to linearity this is true for every size of account.

How on earth would you come up with 9.5 % ??

The yearly reward pool is 75% of the inflation (about 9.5%) of the total amount of Steem. So if everybody would just self-vote (which most do), and nobody would vote on an abusing whale, a whale would only get his/her share of the reward pool proportional to his/her vested Steem back, which sounds much in absolute figures but compared to their Steem Power it is not so much. I think it is much easier for a minnow to make a 100% ROI than for a whale. Whales already own 80% of Steem so it is impossible for all whales to make a ROI larger than the reward pool which is approx. 9.5% per year. An actual analysis of the ROI of whales would be very interesting. I guess their ROI in percent figures of their vested Steem is way lower than expected and more closely to the 9.5% if not less.

Ok. This makes much more sense.
I thought you were retarded, sorry.

I guess their ROI in percent figures of their vested Steem is way lower than expected and more closely to the 9.5% if not less.

smartass @jerrybanfield has done the math and it's more than 100% roi annually.

But that all aside, why would any shareholder recieve any posting rewards in the first place ?

That already is where the abuse begins.
This is a social media blockchain.

Posting rewards should go out to the best content and the curators get curation rewards.

Other than that especially whales should be encouraged to vote on good content so that it attracts better content, so that their stake becomes more valuable.

I think it is much easier for a minnow to make a 100% ROI than for a whale.

The only reason for that being, that there are still people out there (like myself) trying to find and reward good content, but by doing that I will lose out and my stake will not grow as fast as that of a selfvoter.

At the end of the day, no matter how large my stake here gets, if the best way to earn on this platform is by abuse, then my stake will be as worthless as this discussion.

@jerrybanfield show me a whale that makes an annual 100% ROI of their Steem Power, he/she would be a genius :D

But that all aside, why would any shareholder recieve any posting rewards in the first place ?

This is a good point, the underlying reason is that votes are weighted by Steem Power, which is a good or bad thing, depending on how you see it ;)

This is quite a philosophical debate I guess, coming back to the reward curve my initial thought is that n^2 would not really solve the "issue".

Other than that especially whales should be encouraged to vote on good content so that it attracts better content, so that their stake becomes more valuable.

100% agree, at the end of the day I want steemit to be a successful social media network and a successful crypto currency :)

Well then check out @lichtblick.
Posts flowers everyday, trades votes and has one of the highest reps on steemit by now.
He even agrees that his content is shit and he just posts something, so he can vote on it.
Easily made more than 100% ( in vests ) since hf18.
Trust me: n^2 could have potentially solved this issue.

If you don't make it a philosophical question whether posting rewards should be for good posts or for stakeholders, then it's pure game-theory;

Posting rewards for stakeholders without good content would be abuse, because it tries to play the rules.
The solution also comes through game-theory: n^2.

If you don't trust me, listen to @dan who invented this whole thing here and basically writes the same.
I am not as smart as him, but came to the same conclusions.

I've highlighted some of the inherent issues with the linear reward curve in a comment above. I invite you to take a look at it. Feel free to ask question if there is something that isn't clear.

Will n^2 basically increase the reward of posts with a high number of votes, and decrease posts which only received 2-3 votes for example including self-votes? So basically things on trending with hundreds of votes would receive even more, while only self-voted posts will receive less. What if cheaters simply create multiple accounts / bots to profit from n^2 even more than from a linear curve? Suddenly account splitting/delegation would become more profitable than now, wouldn't it?

Its important that we talk about things like the "unfair"reward system. I just made a post about a similar theme, the reward timeline, which in my eyes is complicity for this contraproductivity.

It would certainly have a fun impact on the voting bots..
10 bots with 100SP would be as strong as one 10000 bot, if my assumption is right
:)

You're wrong. First you're calculating two unequal stakes. 10*100 < 10,000. And under n^2 10,000 would be even more powerful than 1,000 compare to 10,000. I've explained some of the math above.

Yes, @felixxx tried to explain that to me already.. It seems, I need a bigger flowchart;)
Maybe I need to see the calculation formulas, or a graph .. or a 1.5h long documentary to understand it better :)

And thanks for the other upvote!
Wouldn't be necessary, as even your un-vote was way more than I expected!
Cheers!

That was a 2% un-vote. I upvoted with 2% too to try to make up for my mistake.