#steem

Reward Curve Deep Dive

reward curve thumb v.jpg

Hello Steemians, I’m @vandeberg, Senior Blockchain Developer at Steemit and today I want to do a more technical deep dive in to the rewards system of Steem and shed some light on the nuances of changing a reward curve. We recently proposed changing the reward curve to a "convergent linear rewards curve" as part of the effort to improve the economics of Steem. What does that mean?

To read more about how the Steemit team believes the economics of Steem can be improved, check out this post on @steemitblog.

ELI5

When Steem was first created, rewards were distributed based on a “superlinear” rewards curve. That meant that accounts with a lot of Steem Power had more innate influence on a per-Steem-basis than smaller accounts. That curve was very effective at encouraging people to consolidate their Steem into one account, which made it easy for other users to find and respond to bad actors. But a side effect of this curve was that the “rich got richer” at a rate which made the ecosystem feel extremely unfair.

Switching to a linear rewards curve meant that every account had the same degree of influence (on a per-Steem-basis) on the rewards pool regardless of the total amount of Steem they had in their accounts. At the time most felt that this dramatically improved the fairness of the ecosystem, and since that change was put into place, we have seen a reduction in the income inequality in Steem. One problem with this change was that it reduced the incentive to consolidate one’s stake in a single account, which made it easier for bad actors to divide up their stake and hide their activities.

Reward Mechanics

To gain a deeper understanding of how these different curves function, let’s look at a few hypothetical scenarios and see how they play out based on how much stake the users have, and the type of curve that is in place.

When a comment is 7 days old, it is rewarded based on the number of reward shares (or rshares) it has, modified by some function called a rewards curve. Steem currently has a linear rewards curve, which means the rshares are modified by the curve f(x) = mx + b. Specifically Steem uses the curve f(x) = x where the rshares are not changed at all. This is compared to a moving total of rshares recently paid out. Let's work through a simple example.

Let's say that Alice authored a post that received 10 rshares. The pool of recent rshares contains 1000 rshares, and the reward fund holds 100.000 STEEM. The reward curve is applied to Alice's 10 rshares, giving us 10 rshares and those are added to the total recent rshares, giving us 1010 rshares. Then Alice is rewarded with 10/1010 of the reward fund, or 9.900 STEEM. The actual values on the blockchain are much different than this, but the math is the same. In addition, the recent rshares are decayed slightly each block. Over time it reaches a pseudo-equilibrium which normalizes rewards over time.

With the same constraints, what if Bob had a comment rewarded at the same time, but he only had 5 rshares. Both of their rshares would be added to the total recent rshares for a value of 1015. Alice would be rewarded with 10/1015 of the reward fund for 9.985 STEEM and Bob would be rewarded with 5/1015 of the reward fund for 4.926 STEEM. As you can see, Alice's comment had twice as many rshares voting for it and she received twice as much STEEM as Bob. We might also call this a proportional rewards curve because the rewards are proportional to the number of rshares voting for a piece of content.

The Different Curves

When Steem was first launched we had an n^3 reward curve, which was changed to n^2 before the first payouts began on July 4th, 2016. Let's run the example again but with a rewards curve of n^2. To make the values somewhat close, we also need to increase the size of the recent rshares to 5000. We have to add an extra step in the calculation that was implicit before, the application of the rewards curve. Alice's comment has 10 rshares, but when we apply that to f(n) = n^2 we get an actual value of 100. Doing the same to Bob's comment gives us a value of 25, for a new total recent rshares of 5125.

Now, Alice is rewarded with 100/5125 of the reward pool, or 19.512 STEEM, whereas Bob is rewarded 25/5125 of the reward pool, or 4.878 STEEM, four times less than Alice! We call this a superlinear reward curve. Certainly this is not fair. Steemit and the witnesses agree, which is why we changed the rewards curve to linear on June 20, 2017 in Hardfork 19.

Superlinearity

However, there is a really nice property of a super linear reward curve. It encourages consolidating stake in order to maximize rewards. In the Steem Whitepaper we refer to this as "anti-sybil". The scenario that it seeks to combat is a single entity spreading their stake over many small accounts in order to hide their actions in noise. A single entity doing this can hide their intentions and siphon off small rewards over time and be difficult to detect. Superlinear rewards incentivizes consolidating stake or at least having the smaller accounts act together, which makes the behavior much more difficult to hide if the entity wants to act in an optimal manner.

Best of Both Worlds?

Is there a way for us to capture the fairness of linear rewards and keep the anti-sybil benefits of superlinearity? We believe so with what we have termed the convergent linear rewards curve. This curve is of the form n^2 / (n + 1).

Let's look at the example above again, but add in Charlie who has a comment with 20 rshares. We will go back to a recent rshare value of 1000. We will also be using the specific reward curve of n^2 / (n/5 + 1) for this example.

Alice has 10 rshares which are augmented by the rewards curve to be 33 rshares. Bob has 5 rshares which are augmented by the rewards curve to be 12 rshares. Charlie has 20 rshares which are augmented by the rewards curve to be 80 rshares.

The payouts then for Alice would be 29.333 STEEM. Bob would be 10.666 STEEM. Charlie would be 71.111 STEEM.

Looking at Alice and Bob, Alice only has twice as many rshares as Bob, but was rewarded with 2.75 times as much STEEM, an increase of 37.5% STEEM per rshare.

However, while Charle also has twice as many rshares as Alice, he was only rewarded with 2.42 times as much STEEM, an increase of 21% STEEM per rshare.

That is the beauty of this curve! As more rshares are awarded to comments, the marginal gain (percent STEEM per rshare) decreases. However, the actual STEEM reward per rshare still increases. If it did not, we would see Charlie rewarded with less than 2 times as much STEEM as Alice, resulting in a negative percent STEEM per rshare.

We have decided to call this a convergent linear rewards curve, because as the number of rshares increases, this curve converges on linearity and behaves more and more like a linear rewards curve. The specific curve it behaves like can be derived by calculating the limit at infinity. In the case of this curve, it behaves like the curve f(n) = 10n. We can tune how quickly the curve converges to the equivalent linear rewards curve by changing the coefficient of the denominator term.

reward curve graph.png

We can graph the derivatives of the reward curves to visualize their "fairness" over time. The red line is n^2, the blue line is n, green is n^2 / (n + 1), and pink is n log( n ). The pink line was added as another point of comparison because it is a common function that grows faster than n but slower than n^2 and is a reasonable candidate for a superlinear curve. This graph confirms what our intuition is regarding these curves from the previous examples. n^2 continues to get more and more unfair as rshares increase and n is fair regardless of rshares. The green curve for n^2/(n+1) is cool because you can see it start unfair with n^2 and end fair with n. n log(n) continues to grow forever. While it is not as unfair as n^2 it does continue to grow in unbounded unfairness, while n^2 / (n+1) does not.

The goal of this post was to deep dive into one of the critical pieces of the Economic Improvement Proposal. Let me know if you have any questions in the comments section below, or what part of the proposal I should “deep dive” into next.

Vandeberg

vandeberg post signature.jpg

Comments · 56

@sisygoboom(65)· 2534d
Hi @vandeberg, asking as a steem developer, it would be fantastic if you could provide us with an algorithm to calculate an account's upvote value with this new curve as the linear equation is now way, way off.
@xyzashu(75)· 2553d
It's not so easy for my math

Alice would be rewarded with 10/1015 of the reward fund for 9.985 STEEM and Bob would be rewarded with 5/1015 of the reward fund for 4.926 STEEM.

For me Alice's figure should be 9.852 instead of 9.985.

And here:

The pool of recent rshares contains 1000 rshares, and the reward fund holds 100.000 STEEM

I need to adjust the figure of 100.000 STEEM to 1000.000 STEEM to understand all the calculations done in hypothetical examples here.

I wonder how everyone else here is understanding these figures. Am I missing something?
@devske(35)· 2556d
Very nice post there buddy. That helped me a lot of understanding how it works. Feels good that the Steem Blockchain is getting a little more fair for everyone.
@kharma.scribbles(58)· 2557d
I will have to read this over a few times, math was definitely not my strong suit.

It sounds to me like the more votes a post gets, the less value each additional vote will hold?

I really hate math lol
@a-non-e-moose(56)· 2640d
I've always thought that policy, especially economic policy, ought to be more directly informed by reasonable mathematics.

We don't see enough of that in government and so forth, in my opinion.

It's nice to see that here on Steem we can use math of such prudent simplicity to so directly influence policy and profoundly impact user outcomes.
@mattockfs(61)· 2653d
Hi @vandeberg , could I ask you to have a quick look at this post where I try to explain why I feel it important to at least use a reward scaling curve that scales with account size, and ideally use a reward scaling equation that results in jumps at fish-size thresholds rather than using a smooth curve.
@jrcornel(82)· 2653d
I can't possibly see how these changes will attract more people to steemit.com and think they are more likely to run what few users we have left away.

I am against these changes as I would much rather get SMTs and communities done first and let them each have their own economic systems.
@redthunder90(52)· 2655d
This explanation is very accurate, thank you very much to let us know how things changed and all the dynamic behind it.
@freebornangel(72)· 2655d

Certainly this is not fair.

Games have rules. If we tilt those rules to favor one side over the other the game changes.

As a user that started with 3 steem, i was best served by the math under the rules of the experiment. I think steem is best served by mass adoption. Capping influence at 500mv (slowly ascending) grows the distribution most effectively, imo.

There isn't as much question if steem will persist at this point. I don't know the actual numbers, but tanking the price is hardly in the maximizer's interest, imo. There are plenty of tokens on steem engine to soak up some of the liquidity created by dumping whales. If they prefer to leave steem entirely, good riddance. Better a lower price than life on our knees, imo. They can only dump it once.

#m2c
@spectrumecons(75)· 2657d
Hi @vanderberg. Thanks for your explanation and applied examples. It certainly makes the theory a lot easier to understand. I have not played with the formula n^2/(n/c+1) before. I think I probably should.

I have 2 concerns based on my early observations.

First is one getting the value of 'c' correct. The low values do not appear to make much of a difference. As 'c' increases to infinity we end back at n^2.

My second concern is regarding rewards posts at the low end. The posts that get the least rshares will take a heavy hit and the posts that get the most rshares will gain very little percentage wise. I am concerned new users will have a hard time getting off the ground.
@rycharde(64)· 2657d
I hope the next post will be on the proposal for 50% curation rewards. Hopefully with illustrations of how it will affect the bidbot economies from the POV of both delegators and bidders. Also, how this will affect curation rewards for upvoters.

BTW it is fairly obvious that something such as the bidbot/vote-buying market will drop in turnover because of the self-vote drop from 75% to 50%. This may, in itself, be the primary reason for implementing this, however, it will not affect the profitability calculation of the individual as it is based on the difference between self-vote and vote-buying.

I look forward to seeing a good reason for why you think this is a good idea.
@dana-edwards(75)· 2657d
And finally and this part is important, but the reasoning to buy Steem? If buying Steem doesn't result in buying influence in some way (if the whales don't actually get more voting power) then how do you convince people to actually buy Steem?

Currently the price of Steem to BTC has been falling off a cliff for some time. I'm not saying it's the fault of the reward curve but if you're going to focus on economics shouldn't you also put some discussion into how you will increase demand for the actual customer (the Steem buyer/Steem investor)?

Fairness for the poster is nice and all but at the end of the day it's the sentiment of the Steem holder which is being neglected the most. Whales are whales because they either bought a lot of Steem or they earned it. I do think yes in 2016 you had a lot of whales who lucked out by being miners or whatever but there are whales who actually bought or earned their Steem only to watch it fall off a cliff.

What do these whales who bought or earned their Steem, and who held, get in return for their investment in the community, the ecosystem, etc? If the answer is nothing then this explains the price of Steem.
@dana-edwards(75)· 2657d
The only way it can be fair is to be able to guarantee that people can't simply buy votes from bots. You have to be able to guarantee that the curation (proof of brain) is functional.

And since you can't clearly guarantee that part why would we be confident that it's going to be fair if every account has equal weight when the majority of votes come from bots etc? How do you change the situation to where posters are encourage to earn votes rather than buy them?
@dana-edwards(75)· 2657d

At the time most felt that this dramatically improved the fairness of the ecosystem, and since that change was put into place, we have seen a reduction in the income inequality in Steem. One problem with this change was that it reduced the incentive to consolidate one’s stake in a single account, which made it easier for bad actors to divide up their stake and hide their activities.

Was the goal "income equality"? I think income equality doesn't create an incentive to make increasingly better content. That said even if there was some level of income equality it would only make sense if the income were substantial enough and with the chance of a great post being discovered, going viral, and getting dramatic income.

Currently just about any post gets around $1-5 no matter how good it is, or what it's about. So now we have no ranking of content quality and no correlation of the size of the reward with the quality of the post.

Fairness might feel nice, but it's never going to be perfect equality. In this case trying to achieve fairness broke the economics for everyone.
@dana-edwards(75)· 2657d

Switching to a linear rewards curve meant that every account had the same degree of influence (on a per-Steem-basis) on the rewards pool regardless of the total amount of Steem they had in their accounts.

And this was the key mistake I think. I think yes we had the problem of whales before, and no system is going to be perfectly fair, but now we have a much worse problem of "buy your votes" replacing "earn your votes".

In other words now only the rich can earn at all because you essentially have to buy your votes.
@stuffbyspencer(62)· 2657d
Thank you for walking through this explanation in a way my little pea-brain can understand, I appreciate it & feel like I might be a tiny bit smarter now! :^)

Posted using Partiko Android
@donkeypong(76)· 2657d
Thank you for explaining this to us. Unfortunately, stake -based voting has not worked. Making some tweaks will change and perhaps improve the dynamic slightly, but in the end it will reach the same point: people will adapt and find ways to honor their own self-interest. Most people do not want to curate or use their downvote power responsibly; they want to make money. Even the most idealistic and long term thinkers eventually realize that they can do a lot more of those things with their own money than they can by struggling to get the system to fit those ends.

Rather than guessing as to what might work in our one giant petri dish, it's probably better to acknowledge that Steem has become the reward layer for investors. Let SMTs create a second layer where much more experimentation can occur. I suspect the models that ultimately succeed will be those that do not try to combine investor incentives with content voting based upon stake, as the objectives tug in different directions.
@edicted(83)· 2657d
I am deeply concerned that this community thinks the problem is that the system isn't complex enough. Complicate payouts and who wins? The people with the most stake, because they have the highest incentive to figure out the new system and game it while everyone else gets fleeced.

No one seems to bring up the fact that Steem's learning curve is already too high for new users. Doesn't seem to stop anyone from wanting to change everything.

The curve you've proposed takes money from the poor and gives it to the rich. It incentivizes Sybil attack. Anyone with 1000 accounts can upvote each account a tiny bit and they'll receive even better rewards than linear.
@arcange(79)· 2657d
Congratulations @vandeberg! Your post was mentioned in the Steem Hit Parade in the following category:
- Comments - Ranked 2 with 128 comments
@themanualbot(72)· 2658d
Simply put, as your rshares increases, your piece of the pie increases that it reaches to certain point of equality or close to equilibrium of your rshares.

With this experimental reward curve, everyone will have their piece of the pie. This will also encourage users to be frugal with their upvotes, the less yet active you are in the platform, the more rewards you reap and more rewarding to contributors.
@trafalgar(73)· 2658d
Thank you so much for being receptive to my ideas of economic reform. I had always been inquiring about whether a superlinear curve with a linear tail was possible and the n^2/(n/c+1) you've posited seems to do the trick very well.

Not to be too pedantic but

The specific curve it behaves like can be derived by calculating the limit at infinity. In the case of this curve, it behaves like the curve f(n) = 10n

If you were referring to your example of n^2 / (n/5 + 1), did you mean f(n) = 5n? As that's where I believe it converges

Now as an alternative, is it possible to simply apply a x% 'spam tax' up until a certain point then allow it to receive full rewards beyond that point? So for example, anything under 1 Steem is taxed at 50%.

Assuming we increase curation to 50% and that indeed becomes the dominant form of staking rewards, it would mean micro votes are earning no more than mean curation and they're probably at least as well off doing something that actually adds value.

A solution such as a tax avoids overly punishing genuine interactions at low levels of rewards such as organic comments while still providing a very clear disincentive for profit based micro vote farming.

Of course overall I'm quite receptive to this curve and think it's pretty neat. I'd like to hear your views
@tsnaks(68)· 2658d
The people with more Rep should have a bigger say in the community
@vikisecrets(80)· 2658d
Thx for the great explanation, I always advocated that n^2 is extremely unfair and preferred linear, also because of its simplicity and transparency. "convergent linear" would be interesting to test, if it really solves some of the mentioned problems. If it doesn't then I would prefer to revert back to linear after some test period.
@svamiva(61)· 2658d
Nowadays if Alice, Bob and Charlie do want to experiment with just any possible rewards curve all they have to do is to publish their posts making the @IloveSuperLinear account 100% beneficiary. Then the off-chain calculations could be made to redistribute rewards between Alice, Bob and Charlie using any desired curve or just using the winner takes it all principle for that matter. However it looks like Alice, Bob and Charlie when they're pushing for superlinear rewards curve do not want to redistribute rewards between themselves, instead they do hope to be able to redistribute rewards from Joe into their own pockets, while Joe might be quite satisfied with the current situation.
@valued-customer(75)· 2658d
I greatly appreciate your cogent discussion of rewards curve, as this is not well understood by many. However, I don't think any rewards curve solves a basic problem that prevents rewards from inuring more to higher quality posts. While non-linear rewards curves may reduce Sybilline issues, they do not discourage extractive profiteering, simply varying how profitable it is. Non-linear reward curves so far discussed all increase the potential rewards for self-voting, delegating to bidbots, and other essentially fraudulent mechanisms to extract rewards that are intended to be delivered by curation, which is intended to best reward the best content and thereby create incentive to create content of high quality which in turn attracts eyeballs to the Steem blockchain and investment by new users, thus producing upwards price pressure on Steem and creating capital gains for stakeholders.

I joined just prior to HF19, and since then Steem has dropped about 30 positions in market cap ranking to around 60th today. I believe the reason Steem underperforms the market is because of the incentive rewards create to extract rewards from the pool, rather than curate good content, as this is clearly profiteering (extracting operational resources) and the opposite of investment (which seeks to generate capital gains). I don't see how altering rewards curves, or increasing curation rewards - which are just additive to author rewards as mechanisms for profiteering, and essentially ignored by folks without substantial stake - can change these incentives.

However, there are mechanisms that can. There is no mechanism that encourages delegation or direct funding of development, and recently the @steemalliance has officially elected a structure by which to operate means of choosing development operations. Also the SPS was completed by @blocktrades to provide mechanisms to deliver that funding recently. Were dividends potential to stakeholders for delegating to or funding directly via SPS the development intended to increase the value of Steem, and thus create capital gains, real investment would be encouraged. I strongly recommend some mechanism be provided that creates financial incentive to invest in Steem, rather than extract rewards better intended to market Steem to potential investors.

Additionally, rewards are currently unlimited and this creates a vector for profiteering. I propose an algorithm be applied to author rewards that confined rewards to no less than 3% nor more than 300% of the median reward, and eliminating curation rewards altogether. Curation rewards don't create incentive to choose high quality content, but instead are simply gamed by substantial stakeholders to maximize rewards. This is not curation at all, but merely additional extraction of operational resources, and only of import to folks with substantial stake.

By limiting curation rewards to a multiple of the median, most profiteering would be eliminated, as the potential returns would be insubstantial for profiteers seeking large returns on large sums. However, by leaving two orders of magnitude available to differentially reward content of lesser or greater quality, significant incentive to produce higher quality content is preserved, albeit not unlimited. Coupling this with a dividend mechanism that potentiates substantial returns from substantial funding or delegation to development restores the utility of rewards as a mechanism to market Steem, as well as introduces another to drive capital gains, and eliminates the negative pressure profiteering exerts on Steem price.

I am not particularly wedded to these ratios, nor specific mechanisms to create these healthy incentives, but reckon these will do the job of attracting investors rather than profiteers to Steem and create capital gains. Real investors will respond to appropriate incentives and the high quality of the Steem blockchain, use case, and community has very real value that has heretofore not produced nominal capital gains. Nominal incentives is how to change that, IMHO.

I'd appreciate your thoughts on these matters, as you are clearly of agile mind and competent to consider how these changes would affect Steem, Steemit, and capital flows (although that may not be your particular area of expertise).

Thanks!
@mattockfs(61)· 2658d
Maybe consider an alternative more easy to understand for regular users.

We have the redfish/minnow/dolphin/orca/whale scale.

Just do something like this:
- Redfish: n
- Minnow: n x 1.025
- Dolphin: n x 1.05
- Orca: n x 1.075
- Whale: n x 1.1
You could easily turn this into a curve,

vs

, where V denotes the accounts vesting power, but that would defeat the purpose of being easy to communicate. You could just sell it as 2.5% fish size bonuses and everyone will understand.
@revo(62)· 2658d
Thanks for this post. It's great to get some examples and explanations of their pros and cons.
@inertia(75)· 2658d
I think the only flaw is:

... if the entity wants to act in an optimal manner.

I'm not convinced any curve will dissuade an entity that doesn't want to act in an optimal manner. And I'm not convinced that the people we're talking about adhere to the description of "optimal."

Meaning, there are plenty of people who will act out no matter what the curve is. I was happy with whatever curve was in play. I'm happy with the current "curve." I'll be happy with whatever replaces it now. I do not see the curve as a magic bullet to get bad actors in line.
@vimukthi(76)· 2658d
Excellent explanation and for the first time I'm totally OK with a non-linear curve. Thank you for your hard work and explanations.
@cardboard(65)· 2658d
By the way, does this new way means that small authors will receive less payouts?
@cardboard(65)· 2658d
You know what would be cool? If you guys would create a tool that would allow normal users to compare payouts using both (current/new proposed) methods :)
@marki99(62)· 2658d
I do not think this can help solve PoB. 50/50 curation and a separate downvote pool might.

Also, we should maybe add some centralization on steem, demanding from bidbots to burn 10-20% or the steem they earn. This will still allow investors to earn a good return while helping the price of steem.

Eyeballs will help the steem price raise. Advertising through bidbots is an amazing sink for the coin.
@lordbutterfly(80)· 2658d

which made it easier for bad actors to divide up their stake and hide their activities.

Who is hiding their activity? This is such a non argument that it makes the whole post look silly. There is no one hiding anything. What delusional parallel universe are you people living in? Everyone knows who everyone is and what and why they are doing what they are doing. The onchain action delivers information and reveals motivation.

You overly complicated the text but if read carefully the context is extremely simplistic. It all returns to the original quoted sentence that contains an argument that is extremely unimportant and doesnt justify any change even in the slightest.

Jesus Christ, high stake holders will do whatever they want when ever they want to do it and you cant do anything about it. No one will do anything about it.

Talking about stake divide and someone "hiding their activity" is a discussion for the deluded or the ignorant that have no clue what steem interactions look like.
@stackin(78)· 2658d
This may work but more things needs to be added.

This is a suggestion i wrote on steemit blog post...

What if we limit a person to 1 upvote to an particular account for 48-72 hours. This may spread many people to upvoting and curating better content/accounts.

There are other platforms that are currently doing that and it’s making the curator vote for others as they can’t keep voting the same person all the time with a three day period time limit 🤔

This will tame the bidbots, selfvoting, and circle dick suckin' action LOL
@justineh(72)· 2658d
Great explanation, thanks for taking the time to do it.
@jaki01(77)· 2658d
Interesting!

My own idea, which I formulated about a year ago, is a reward curve which started as a*n^2 (thus flat at the very beginning and then getting steeper), and then later changed into linear which would work against self-voting as well as against the excessive rewards of pure n^2. I see your curve would end even flatter which I think is reasonable.

@clayop had a similar idea.

The curve in my mentioned post could be made more 'beautiful' (without a 'knee') by using spline interpolation, but it was about the idea (the mathematics would be your part then :-):

I also take the opportunity to present my other idea of implementing diminishing returns when upvoting the same accounts (including one's own) again and again to make it more attractive to upvote many different accounts instead of always the same 'best friends'.
@tsnaks(68)· 2658d
I hope I am not the only one left super confused about this
@holybranches(60)· 2658d
This should have been used in the @steemitblog post. Especially the graph part.

Add this to the featured posts ASAP.
@discernente(77)· 2658d
I think revenue sharing would be an important step forward for the Steem Economics and it would encourage people to power up even more.

Posted using Partiko Android
@revisesociology(82)· 2658d
The pub's not the best place to be reading this but this does seem like a reasonable compromise as it disuades people from voting on low value 'dross' and should encourage votes away from overvalued 'dross'.

Yes there's a lot of things it won't solve as most of the early comments here point out - but focussing on this new green curve all other things being equal this seems like it should encourage both curation of higher quality posts and wider distribution of curation while dusuading vote farming.

Glad this is back on the agenda.

Posted using Partiko Android
@martie7(62)· 2658d
It's not so important now. Better to make RC delegation system. Steem already very popular platform but we need free registration system and good scalability.
@masterthematrix(61)· 2658d
I would like to see, that the Coin "Steem" and its "Reward Pool" will be designed more towards an infrastructure Coin instead of using it primary for POB content discovery. The reason for that is, that with the introduction of communities and SMT the Steem Ecosystem is not only more about "Content creation" and discovery. No, we have a bunch of financial dapps, gaming dapps, services etc... which don't have anything to do with content discovery. All those Dapps should have more or less "equal" chances to receive Support from the underlying Steem Reward Pool mechanism to earn and empower their communities. I think Steem is good with Steempower for Voting on Witnesses and Proposals (plus a new feature in the future Voting for Communities and Dapps) and as a source of Ressource Credits. For me that should be enough, on the level what Steem is doing to support its Ecosystem and for the use of the POB mechanism...only vote for Witnesses, Proposals and SMT or Communities. No, more voting and abusing the reward pool by creating and discovering or gaming the content creation part. All this POB content creation and discovery mechanism should be happening on the level of SMTs, and its communities, because these groups will have Admins who are able to set in rules how each community handles its users and the way they value content creation. If Steem gets understood as a infrastructure Token and SMTs are understood as a empowering Community Token, than we have a clear Vision for all participants Investors, Users and Creators alike. I don't like the idea that we are aiming for a SMT Ecosystem where all these project fight for attention but the real mining of Steem only happens on the content side of things...somehow that doesn't make sense. Why should other Dapps or Communities suffer from miss management of the Steem Reward Pool which is only used for content creation and discovery?

Posted using Partiko Android
@isb(65)· 2658d
steemit is not paying already a good reward to investors only 1% per month on their holding which is very low on other hard the value down and down and people here instead of making money lossing more so this new system if it decrease the reward they discourage investors more and they will power down and leave steem forever
@oldtimer(76)· 2658d
After reading this I will never ever dare to comment on this issue. Keep doing a good job, I'll crawl back to my little cave and put some wood to keep my fire alive.
@ew-and-patterns(71)· 2658d
Thank you for breaking this down in understandable language!
@bashadow(70)· 2658d
That is fine, and it may fix peoples idea of entitlement, but it really honestly has absolutely nothing to do with curation. This will not fix curation. It will not help the good content to rise to the top. All this does is fix the greed of entitlement.

Curation is about selecting post a person thinks is of high quality in their eyes and thus they vote for it. They do not time their vote, they vote when they read and see it. The above does nothing to curb vote bots, or vote buying, or vote trail activity. It changes nothing, and fixes nothing other than the greedy feeling they deserve more rewards for their votes.

If you want to cure what ails the curation issue, perhaps looking into randomizing the start time for the curation reward pay out. Yes it ends on day seven, but you can fix the formula to where the payout schedule starts at any point along that 7 day window, not just at the start of when the post was posted.

That would negate most of the bot activity and selling of votes, because someone with a smaller account may get more curation rewards than someone with a larger account, so it is not likely to happen.
@d-pend(76)· 2658d
First, I'm excited to see how this will go because I've wanted to see how something other than pure linear would work in real-time (wasn't around back in the early days of superlinear.)

However, regarding the curve "converging" on linear at the top, won't this just encourage people to split their stakes into whatever quantity is optimal return (before it starts to flatten) or am I conceptualizing this incorrectly?
@joshman(74)· 2658d
Are developers really good at anticipating human behavior? Not the ones I know!

Posted using Partiko Android
@prydefoltz(75)· 2658d
Thank you for the illustration of the derivative work in the rewards pool. I am not sure that changing the formula however would get the results we want: the growth of the block chain, more users, more ad revenue, and a higher value for steem.

I would like to see a better balance between rewarding those who do the creative and supportive work for the block chain and those who invest but mainly reward themselves.

I would like to see a minimum vote power instituted for those steemians in good standing. For example, everyone gets say 10 one cent votes a day to use but they cannot use it on themselves. This would encourage users to network, use the block chain more, and make newbies feel they have more impact on the platform. Now that we are an ad-revenue based platform/block chain, we need this.

Some of that ad-revenue, once running costs are met, could be used to reward those steemians that perhaps are not as heavily invested in steem but have invested considerable time and intellectual property building the block chain. Better still if they happen to be both invested intellectually and monetarily.

We really need to have our view counts back. Right now the only reliable evidence of proof of brain visits is the comments section. We need to encourage the use of our platforms and reward those who use it as intended. More eyes of the page will drive up the value of steem and rewards for everyone, no matter how we divvy it up.
@bluerobo(70)· 2658d
Let's not get side tracked now and push SMTs out the door first.

Posted using Partiko Android
@markkujantunen(77)· 2658d
I'll copy a comment of mine I posted to a blog post of @whatsup's that discussed this proposal. The quote is @whatsup talking.

Let' build communities and SMTs and attract new users, let's not promise them riches, or content discovery, let's tell them to build a social platform in which they can have partial ownership.
This.

I agree on any tinkering with the reward distribution math almost certainly not helping if the goal is to improve content discovery or stake distribution. Projects like @ocdb are much better at stake distribution. It all comes down to finding an acceptable compromise between rewarding stakeholders and helping newbies grow in order to make the DPoS consensus mechanism more secure. Judging by the 4.1 million SP delegated to @ocdb, a lot of stakeholders think getting 90% as much as by self-voting is enough and giving up for 10% to altruistically further the common good is ok.
@aussieninja(70)· 2658d
To be fair, I don't super amazingly understand it all, but the green curve does seem preferable.

Bad actors are always going to find a way to be bad... and everyone has their own agenda so getting consensus on these types of things is potentially impossible... especially something that may improve things for smaller accounts but not benefit larger accounts like the witnesses.

For the most part though, I think whatever curve helps newer accounts feel at home is probably preferred. It's classic game theory, if newer people can get rewards then they'll come back... and basically maxing out the rewards larger accounts can get from the reward pool and redistributing that to the smaller, newer accounts, the better for the platform.
@clumsysilverdad(65)· 2658d
Thanks for taking the time to put this out there and share the info. Kudos
@blockchainstudio(72)· 2658d
nice! but imho, ad-revenue sharing with authors is much more important, and i'm not sure if steemit is even considering this. decentralized PoB is (almost) impossible (by the simple prisoner's dilemma). Nothing changes for people who use alts A and B to vote each other no matter which reward curve or author:curation ratio are used.

of course, i like the idea to try new things including this and reward ratio. but i sincerely hope that Steemit also admits that without ad-revenue sharing and centralized curation (e.g., utopian), PoB is impossible.

Google gets more than 100bn, facebook gets 50bn from ads, how Steemit can properly reward really good authors without ad-revenue sharing? impossible.

Thanks.
@josediccus(83)· 2658d
I'm definitely not saying this is not good, but isn't this rather promoting capitalism ahead of the communal aspect which made the blockchain thrive in the first place?
@steemitboard(66)· 2658d
Congratulations @vandeberg! You have completed the following achievement on the Steem blockchain and have been rewarded with new badge(s) :

You received more than 250 as payout for your posts. Your next target is to reach a total payout of 500

_{You can view your badges on your Steem Board and compare to others on the Steem Ranking} _{If you no longer want to receive notifications, reply to this comment with the word STOP}

To support your work, I also upvoted your post!

Do not miss the last post from @steemitboard:

New japanese speaking community Steem Meetup badge
Vote for @Steemitboard as a witness to get one more award and increased upvotes!