My thoughts on the Steem reward curve

in #steem6 years ago

I’ve recently seen the idea floated around that moving to a linear reward curve was a bad idea. I’m writing this post to help open up the discussion and discuss why I think going back to the old rewards curve is a very bad idea.

For those unfamiliar with the reward curve and Steem voting system, here’s some elementary background information on what it all means. When Steem was first developed, the creators wanted to find a way to prevent abuse of the system. If everyone had an equal vote, one could create many fake identities, use those accounts to skew voting, and distribute a large portion of the reward pool into undeserving hands.

Vote manipulation is already a problem on platforms where users aren’t paid, God knows how bad it’d be on a system like Steem. Thus the “One Steem Power, One Vote” system was born. Now, each user doesn’t have the same vote. The power of your vote depends on your stake in the system.

The creators took it a step further to counteract abuse. Operating under the (probably true) assumption that it’s easier to police a few significant instances of abuse than it is to police many smaller abuses, the ^2 reward curve was instituted. There were other reasons for the ^2 reward curve, but I don’t think they’re relevant enough to go into in this post.

What does ^2 reward curve mean?

Math. When we start talking about it many peoples’ brains shut down and drown it out. This concept isn’t that difficult to grasp, so bear with me.

Those of you deeply familiar with Steem will say “your summary doesn’t even mention RShares and this and that.” Yes, absolutely, I deserve that rebuke. For simplicity’s and brevity’s sake I’m going to try to keep things a bit more basic.

^2. Basically, it means that the more Steem Power (SP) that votes for that post, the value of the post increases exponentially.

For example, a post with 1000SP worth of votes will be worth 100X more than a post with 100SP of votes, despite only having 10X the votes. A 10000SP post will be worth 10,000X that of 100SP post, despite only having 100X the amount of votes.

Under ^2, the value of your vote may only be worth $0.01 on an unvoted post, while your vote is worth $2.00 on a post that’s already at the top of trending.

What does a linear reward curve mean?

It bothers me a little that people call this a curve. Linear means it's a line. When it's a line it's no longer a curve. But I digress.

Under a linear "reward curve," your vote is worth the same regardless of how much the post is currently worth. If your vote is worth $0.25 on a $0.00 post, it will be worth $0.25 on a $1,000 post.

What would the value of your posts look like if we went back to ^2?

A $10 post today may only be worth $1. A $50 post today might be worth $10. A $1,000 post today might be worth $10,000. Low-value posts would be worth significantly less, and extremely-high-value posts would be worth significantly more.

What are the arguments for returning to ^2?

To counteract abuse. Today, users can engage in vote collusion to relatively easily acquire a significant portion of the reward pool. Also, self-voting has become far more lucrative.

What about having upvotes on a linear scale and downvotes on a ^2 scale?

I think this idea has a lot of potential. The larger the abuse, the more power users have to counteract said abuse.

Why is it a bad idea to completely return to the old ^2 reward curve?

It will be more difficult to discover great content.

Today, someone with 20,000SP can vote on a new post and increase its value by $5. There are a lot of users with that amount of Steem Power, which means many eyes to spot and upvote great content. Increasing a post’s payout by $5 really helps the visibility, and it only takes a handful of votes from these dolphins to make the post highly visible.

However, if we were to go back to ^2, that same vote from the 20,000SP user may only be worth $0.20. For a post to go anywhere, a massive whale would have to discover that post. Today, considering Steem’s popularity, there are too many posts and too few whales to handle that kind of posting/voting dynamic.

It will be extremely challenging for new users to get anywhere.

This ties into the last point. It’s difficult enough for new users to gain traction and become successful bloggers on Steemit. With today’s number of blog posts to sift through and curate, combined with today’s few number of users with huge amounts of Steem Power (one of whom would have to upvote your post to make it relevant), it would be many times more difficult for new users to become successful than it is today.

Steem Power will become much less valuable and the price of Steem will show it.

How many people are going to go out and buy 1,000SP when it only increases their vote value by a penny? I can’t say for sure, but I’m confident it would be far fewer people than today. How many users are going to Power Down and sell their Steem when they see the value of their vote drop from $5.00 to 20 cents? How many people will throw in the towel and decide that owning Steem Power isn’t worth it anymore?

I think a change back to a ^2 reward curve would substantially hurt the price of Steem because it would significantly decrease the value of Steem Power.

Sort:  

The current system is design to support minnows in Steemit.

Minnows certainly have a lot more power under this system than the previous reward curve.

Reputation should have a bigger effect on flags to fix the abuse. If a reputable dolphin flags something, that should be enough to correct the abuse from non-reputable whales and dolphins abusing the system.

I think that idea has some potential. If reputation plays a significantly larger role, I wonder in what ways would the reputation system be gamed?

I suppose someone would figure out a way to game it. It wouldn't be for any monetary gain though as it should only be for flags and not votes. Probably be exploited for some of these ridiculous flag wars in some way though.

I picture users buying, renting, or making arrangements with the accounts of other high-rep users

this place is amazing... experimenting on the borders of anarchy.

They would have to get reputation actually on the blockchain for it to be calculated into the voting though (as far as I know.) My understanding is that rep is only something through the front-end. But yea..there's always gaming of pretty much any system put in place.

I thought reputation is a part of the blockchain?

It's been quite a while, so I could be mistaken. But I do believe it is separate from it. Though I couldn't easily find anything to say definitively one way or the other.

I'll try to verify with some people and let you know what I find one way or the other. :)

I’d be concerned we’d get into retribution wars like we’ve seen a bit of recently. Someone gets flagged so they then flag the person back on every move they make. I’m not greatly convinced by the flagging system. My rep is too valuable to me to risk a retribution flagging. Not sure how to get around this problem.

The biggest problem with the old system was that it created a very steep hill. To go from no payout to $0.02 was a pain in the ass. Even with a dozen minnow votes, you were still stuck at zero. You had to have a whale or dolphin friend to get into the payment portion of the curve.

And, a problem we still have is that the distribution of SP looks like an exponential curve. So, exponential curve X exponential curve created a steep rise, but also a steep drop off.

Yeah, that was a steep hill to climb! I think curation guilds helped with that - they made the tail of the distribution curve much longer. But at the same time, they distributed a lot of Rshares, which made everyone else's vote worth less. The guilds could find a lot of the good posts, but not all of them. Today, Steemit is far more popular, and the work of curation guilds wouldn't go very far considering the gigantic increase in the number of daily posts.

Good contribution really friend, is very helpful for users who do not move well on the platform is certainly of great importance Greetings I follow you I follow you ..

The old system seems very elitist. I think it would be horrible to go back... thanks for the simple explanation.

I much preferred the old system.
It was like a lottery.
Plenty of posts got very little, but plenty of lottery tickets don't win anything.
People don't play the lottery to make a reliable return on their effort and cost; they do it for that tiny chance at a huge payday.
Whoever decided this place should be fair did a lot of damage.

The curation dynamic was also much different under the old system. Almost the entire reward pool was decided by curation guilds, just about everyone else had only a small amount of power. Around the time we switched from ^2, curation guilds were being phased out, and were far less relevant.

At that time, if your post didn't get a vote from blocktrades, curie, or steem guild, you probably weren't going to earn much on your post. Kind of like a lottery, I suppose. I didn't like that system. At one point I had 70,000 Steem Power and my vote was worth like $0.05.

But what does it do to the quality of the posts?
The closer it resembles a day job, the less authors will devote to finishing touches.
If spending another hour getting the wording and images perfect stands to drive the post from $15-$20, we're going to get a lot less finnessing than if the extra effort stands to take the post from 10c-$200.
It was never about fairness; it was about quality.

I agree that we shouldn't jump back to ^2 rewards, but the issue is that linear rewards are easier to abuse.

Rather than go backwards, we should address the issue. The community disagrees with the valuation of certain posts. Thus, we need to validate large votes in order to protect the rewards pool and only give heavy rewards to those with large community support rather than a single self upvote or a large whale upvote.

I've suggested that we verify votes utilizing a vote verification scheme that compares the large upvotes to ones around it.

Certainly if an upvote is larger than the entire rest of the support for a single post, that is too much.

More generally, these are the approaches we should be making instead of arguing for flawed system #1 or flawed system #2. Pinpoint particular issues, find potential solutions, find flaws in those potential solutions, repeat.

Going backwards won't help the platform move forward.

Let's say any whale has a $300 vote and they use it on another user they like (this happens all over this site), if the total post is worth $315, should the author be entitled to the rewards even though only person is supporting the work to the extent they are?

I argue that such votes should be validated. If @haejin had $300 and only had small minnow votes, then nobody would argue with that, and someone like Bernie would be in the wrong. In fact, some of the tactics that Bernie uses I don't really care for, but the system allows for that so it happens. Hell, some of the posts where Bernie complains are overpriced and clearly don't deserve the rewards they get. Vote verification works both ways. They can also help limit lone downvotes (like the one used on you, right now). That way things are more community based and it will be harder to lone whales to get in these pointless large fights.

That makes no sense, because @haejin earns rewards greater than the amount lost by flagging. Why would he drain the rewards pool in order to enlarge his share of the reward pool? By downvoting @haejin, he wastes voting power. Such an action is unprofitable.

I think there are some issues with the system overall. I wrote a blog post a little while ago talking about cryptocurrencies and the systematic flaws I personally have noticed. I also brought up what I see to be an issue on here, the fact that someone can effectively use their economic influence to "buy" power on here. Though I think the overall system is a great idea, I don't know how far we can get in another system where the richest hold the majority of the power and the content we see. After all, this is something we talk about every day in real life (people buying newspapers and controlling the content, large companies manipulating what they want us to see, etc). Of course, Steem and these people aren't large companies, but I think the idea of the richest controlling the content is the same.

https://steemit.com/cryptocurrency/@mistakencynic/cryptocurrency-and-wealth-distribution-systematic-flaws-part-1

While at the same time the power is in the hands of a relative few, those with the most power also have the most to lose if the system fails. They also have the most to gain if the system succeeds.

The fact that people can buy Steem Power, I don't see as a problem, I see it as a feature. In fact, when someone buys Steem Power, they are benefiting everyone else involved in the Steem ecosystem, because they are the ones who are attributing value to the token. Without people buying Steem, Steem the token ceases to have value in the real world, and what you end up with is a Reddit where you can earn worthless digital tokens.

yeah personally I thought the old reward system was ridiculous and I'm sort of surprised that it is even something that gets brought up to maybe go back to it. It was just ridiculous. We are way better off now.

Bernie used to do all his flagging from one account. It's not like Reddit where 1 account equals 1 vote. You need SP behind your vote for it to matter.

On Reddit, Facebook, Instagram, Youtube, and basically everywhere but here 1 account = 1 vote.

Here, 1 account doesn't equal 1 vote. 1 Steem Power = 1 Vote.

If bernie has 1,000,000SP, it doesn't matter whether he has all his SP in 1 account or spread across 100 different accounts, he has the same amount of influence either way.

"he uses these account to up vote eachother"

Yes, he does. The end result is no different than him upvoting himself with just one account with the same amount of Steem Power.