An Illustrated Guide to Curation - from the simple to the complex - with real examples from past posts - Part 1steemCreated with Sketch.

in #curation6 years ago (edited)

Frontcover.png


Curation rewards are a simple idea. But beneath the surface lie a multitude of complexities. In this series of posts we use illustrations and examples from past posts in an attempt to lay those complexities bare.

In part 1 we cover the basics, including a novel way to visualise each upvoter’s share of curation rewards. We also reveal why the percentage of rewards given to curators is not close to 25% of post payouts. And we explain how authors actually receive curation rewards from upvotes made after thirty minutes.


The Basics of Curation

Curation rewards exist to incentivise readers to seek out quality new content and to promote it with their upvotes. A proportion of the rewards from each post is reserved to pay for this curation. The currently split of rewards is 75% of the post payout for authors and 25% for curators. The overall curation payments are thus higher for more successful posts, rewarding curators who find quality content.

Curation rewards are also higher for early curators than for late curators, rewarding readers that take the time to unearth new content. However, for curation undertaken in the first thirty minutes after an article is posted, a proportion of the curation reward is returned to the author. This aims to prevent an immediate rush of curation on well-known authors. The proportion returned to the author decreases linearly, from 100% immediately after posting, through 50% of the curation rewards for an upvote at 15 minutes, reaching 0% at thirty minutes and for any upvote thereafter.

Curation rewards are always made in Steem Power (SP). This differs from author payouts which may be either fully powered-up, so 100% rewarded in Steem Power, or a 50/50 split between Steem Power and Steem Based Dollars (SBD).


In the simplest possible example, a post which receives a single $4.00 upvote after 30 minutes results in a $3.00 author reward and a $1.00 curation reward for the upvoter. Under the 50/50 SP/SBD author payout option this translates to the payouts below.

firstdiagramv2.png
(in all examples we ignore any price / reward pool movements over the week that could increase or decrease the post value after an up vote is made)

Under the same example, but with the upvote made after 15 minutes, 50% of the curation reward is returned to the author, resulting in the following payouts:

firstdiagramv-15mins v2.png


Depending on the size and timing of upvotes in the first thirty minutes, the total curation rewards on a post can vary between 0% and 25% of the overall post payout. Here are a couple of examples from past posts:

For this first post by @crema, no upvote was received in the first 30 minutes. The curator payout of $1.37 is exactly 25% of the post payout of $5.48. This example, with the full 25.0% obtained by curators, is actually quite rare.

example25perc.png

For this second post by @fivestargroup, a huge whale upvote was received after six minutes, donating a large proportion of the curation rewards to the author. The curator payout of $6.37 is around 6% of the Post Payout of $105.49. This situation is more common, although this example is at the extreme end of the spectrum.

example10perc.png

To give an idea of how curation reward percentages on posts vary, the chart below captures the curator rewards expressed as a percentage of the post payout for all posts on a single day in October. The figures have been rounded to the nearest percentage. Only posts where the author payout exceeds $1.00 have been included.

The average percentage is approximately 18% (by a count of posts, not weighted by post payout) so significantly less than 25%. Most posts receive some upvotes within the first thirty minutes, often from curators but also from the author themselves.

Screen Shot 2017-10-19 at 11.02.16.png


How the Curation Mechanism Rewards Early Upvoters

The order of upvotes makes no difference to the overall post payout. It does not matter who votes first, whether small votes are made before larger ones or vice versa. The upvotes simply add to the post payout, one on top of the other.

If we ignore the complication of the first thirty minutes, the author and total curator reward amounts are also unaffected by the order and timing of upvotes. However the order of upvotes is critical in determining the split of curation rewards between upvoters. The mechanism is designed to reward early curators who take the time to unearth new content.

The curation “weights”, i.e. the proportion of the total curation payout received by each curator, are calculated based on a square root function. The “weight” given to the first upvoter is the square root of the rshares added to the post by that curator’s upvote, where the rshares can be considered as a measure of the size of the upvote made. The “weight” given to the second upvoter represents their addition to the square root of the total rshares on the post, i.e:

curation weight of second voter =
√(rshares added by first voter + rshares added by second voter) - √(rshares added by first voter)

If there were no square root function applied then the curation reward weight for each curator would be in proportion to the size of their upvote. The timing of upvotes would also become irrelevant. The square root function has a lower impact on small numbers; for example compare SQRT(4) = 2, an impact of 2, and SQRT(81) = 9, an impact of 72. The function achieves its purpose by having a small impact on weights when the overall post payout is small and a larger impact on weights when the post payout has increased. This acts to increase curator rewards for early voters.


In this second example we have two upvotes, both of $4.00, the first made 30 minutes after the article was posted, the second one hour later. For simplicity the curation weights have been reduced down proportionately to the size of the curation payments; this does not impact the calculations.

The post payout is $8.00 resulting in a $6.00 author reward and a $2.00 overall curation reward. As can be seen in the illustration, the first upvoter receives a $1.41 curation reward, around two and a half times as much as the $0.59 received by the second upvoter.

2uptoefull.png

The relationship between the curation payments and the weights is represented visually at the bottom of the above illustration. The overall curation payment of $2.00 is the total area covered by the blue square and green angle shape combined (both green and blue areas are equal at $1.00). The curation weights are represented by the distance along the x-axis (√1 = 1.00 for the first curator; √2 - √1 = 0.41 for the second curator). The curation payments are represented by the vertical area above the x-axis, so the blue and green area between the red lines for the first curator (weight 1.00 x height 1.41 = $1.41); the completely green area remaining for the second curator (weight 0.41 x height 1.41 = $0.59).

Expressed more simply, what we can see is that the first curator gains almost half of the curation rewards added to the post by the upvote of the second curator (almost half of the green area sits above the blue square).

Under the same example, but with the first upvote made after 15 minutes rather than 30 minutes, 50% of the first upvoter's curation reward is returned to the author. Note that even though the timing of the second upvote is unchanged at 1hr30 minutes, a proportion of the curation rewards added to the post by the second upvoter is returned to the author. It can be considered as being passed from the second curator to the first curator through the square root function and then from the first curator to the author through the 30 minute rule. The second upvoter’s curation rewards are unchanged between these two situations, the question is whether the curation rewards they add remain with the first curator or are passed to the author.

2uptoefull_50%.png


We can expand this theory to multiple curators and observe the same behaviour. This third example is based on four upvotes of $4.00 made at thirty minute intervals after the article was posted. The illustration below shows how the curation rewards build up with each upvote. Note that the weights along the x-axis do not need to be recalculated with each upvote, they simply form a smaller proportion of the total weight each time an upvote is added.

4blokcs.png

From the illustration above we can see that:

  • Each new upvoter provides some curation rewards to all previous curators.
  • The curation rewards for the first upvoter have doubled from $1.00 to $2.00 since our first example (while the post payout has increased by 4x).
  • The curation rewards for each new upvoter decrease. Later curators receive close to the minimum limit. This is half of the curation rewards added by the upvoter (i.e. $0.50, half of $1.00).
  • If we were to change the time of the first upvote to 15 minutes the author would receive a portion of the curation rewards added by the upvotes of all curators (passed through the curation rewards of the first upvoter).

It is worth noting that it is not the number of previous upvoters that is important in determining the split of curation rewards but the amount of the upvotes added. It would make no difference to the curation rewards of the fourth upvoter if there had been one previous upvote of $12.00 rather than three previous upvotes of $4.00.


To finish we look at a real life example. Five curators, the first being the author with a very small upvote at t=0. A second upvote of $1.48 by @floridasnail after 2 minutes and a third of $4.88 by @bramd much later. Other upvotes are small (but no doubt very welcome!)

Screen Shot 2017-10-19 at 18.31.56.png

The charts show three different curation methods.

  • In blue: Linear curation with no 30 minute rule - i.e. curation rewards in line with upvote size.
  • In orange: Square root curation to show the impact of rewarding early curation.
  • In grey: Square root curation but also showing the impact of adding the thirty minute rule.

As can be seen in blue, @bramd makes the largest upvote and under linear curation would receive the largest curation reward. In orange we see that under square root curation @floridasnail receives almost the same curation reward as @bramd due to voting earlier, even though their upvote was one quarter the size. In grey we see that almost all of @floridasnail’s curation rewards are passed to the author as @floridasnail upvoted after 2 minutes.

On the right hand chart we can see that the author receives $0.72 of curation rewards. All of this comes from @floridasnail’s square root curation rewards (which reduce from $0.77 in orange to $0.06 in grey once we add in the thirty minute rule). However, @floridasnail has obtained $0.35 of this $0.72 from the upvote made by @bramd. Looked at another way, the author receives $4.03 from the upvote made by @bramd, so 83% of the upvote rather than 75%. This demonstrates that a proportion of curation rewards added by upvotes made after thirty minutes can still flow to the author.


That’s all for part 1. In part 2 we will explain why most upvotes receive close to the minimum level of curation rewards and use real examples to examine how feasible it is to receive more than your upvote value in curation rewards. We also reveal why you may not be aware of the most valuable curation you undertake.


Thanks for reading!

footer1.jpg

Sort:  

This is an absolutely awesome post! I've been recently wondering how exactly curation rewards are calculated and could not find anything remotely this detailed with a good bit of Googling.

Not only does it provide all of the detailed information but it explains it in a super easy to understand way. Thanks to the graphics I was able to understand the concept almost instantly without having to get into the actual math.

This should win some kind of award. Seriously.

Now that I'm done heaping praise on this great post - I'm thinking about the value of the curation rewards system overall. I understand the point of it of course, but I think it makes things worse as far as distributing rewards and actually surfacing good content.

In practice it encourages people to vote on posts by popular authors that are likely to get a lot of rewards (regardless of the quality of the post, and often without even reading the post) and disincentivizes people to vote for unknown authors who post really good content since those posts are likely to get very little payout.

I wonder how things would be if there were no curation rewards. I suspect the popular authors would get fewer votes since there is no financial incentive to vote for them, and that would really help spread the rewards around more.

It would be great if there could be some curation system that rewards voting for good content from unknown authors. I am going to think about how that could work. When SMTs come out it will be interesting to see how some of them work out with different curation parameters.

Anyway, again great post. Please keep it up!

Thanks @yabapmatt! Both for the praise and for this great comment! I'm certainly planning to add more parts to this guide. I'll have to find a contest to enter it into!

I'm not sure on your conclusion that it makes the distribution of rewards worse though. I'll be looking at this more in part 2 or 3 but my thoughts so far are:

  • You can obtain curation rewards of 25% of your upvote if you vote on an unknown author who has a post payout at zero (or close to zero and your upvote is more material). There's also a chance for greater rewards if the post does become popular, perhaps through your resteeming.
  • If you vote last on a whale post, or towards the end then you only receive around 12.5% of your upvote in curation rewards (because you give half of your 25% to the people in front)
  • If you vote in the middle of a whale post then the rule of thumb I am working on is that there needs to be three times as much upvoting after your curation than the amount when you curated for you to reach 25% of your upvote (so that the height of the new total curation square is twice the height of the square when you voted). So if you think the post will reach $100 then voting at $25 post payout will get you back to 25% curation. (So only one in four people make it to 25%? - I need to think about that more)
  • If you vote at the front of a whale post before it becomes sizeable then you can make bank. But you are likely to be in the first 30 minutes. In addition if the whale has already upvoted their own post their own vote will be large so they will capture the curation rewards.

I need to do more research and map out a few individual posts to see the best point / type of post to upvote. I'm sure it will differ by author not just by size. And my guess is that unknown authors who produce single pieces of great content will rank above the whales for curation rewards. We'll see!

Really enjoying this back and forth in the comments - something I'd love to see more of here.

I would love to see some data on the curation returns from voting on an "unknown" post vs a popular author. I just kind of assumed that the return would be better voting for someone popular because their posts will earn more but now that you've shed some light on how it works perhaps i'm wrong.

But even if the math works out in favor of unknown authors for curation rewards, I suspect people just assume like me that if they want more curation rewards they need to vote on posts by popular authors at around the 30 minute mark.

When I say it makes the distribution of rewards worse, I mean that because I suspect that people just upvote popular authors to get the curation rewards and don't upvote unknown authors because they don't feel there is much in it for them. If this is the case (and I only have anecdotal evidence) then I definitely think it would help to concentrate rewards with popular authors.

But like I said - can't wait to see any real data you can dig up, and also once SMTs come out we can start to see the effects of other curation strategies.

I should have some data in the next couple of days. I think the results will vary between different authors so I may have to categorise into a few typical patterns even within the whales section.

The SMTs is a great point. I'm just hoping it's not too expensive to set up. I'd like to be able to distribute my own tokens and set my own reward rules!

Reading back old posts can be really good and fun. In this case I even decided to react, although it's a very old post and you probably don't remember precisely where it's all about...

This post was from 6 months ago. Back then you, and probably many others, were already hoping to see some good things from SMT'S. But one again it seems that things go very slowly, perhaps to slowly?

Then I came up with an idea that might help minnows. Why don't we have a system where higher upvotes provides a lower percentage pay out?

I don't know how to call it precisely, so let me show it with an example.
A post with 1 upvotes gets a post out of $1.00.
A post with 2 upvotes gets a pay out of $1.95
A post with 3 upvotes gets a pay out of $2.85.
So the additional $ will decrease per extra upvote. And where I refer to upvotes I actually mean SP/VP.

What do you think?

I also love this back and forth. I think with the rise of human whale curators, the whole system has returned to incentivizing the finding of good content first. Which is still hard because there's just SO MUCH. But if you can find a post that WILL get a curie first, you're the most likely to get a good curation reward. Alternatively, you can figure out the voting patterns of some trails and get in early on those. But that way lies a race to 0%

I am so thankful for this post! Curation always seemed like this thing that nobody was even bothering to explain.

My vote is only worth about .11 now above 90% power where I keep it. I almost never vote on the "popular" posts as you think of them. My best reward comes from a post that pays out between $10 and $60 maybe. When I get on a huge post early, I can do ok, but it's much easier for me to find the littler guys to jump on. I make good rewards if I vote at 50 cents and the post goes to $2 even. It's a curation fallacy to think you have to only get on $100 posts - that's actually worse rewards unless you are one of those top 100 big guys. You get "stomped down" by their big votes.

Thanks for the response @fitinfun...my guess is that most people don't know what you do (I didn't until this post!) So if most people believe the fallacy then it is still a big problem - if not with the algorithm then with messaging.

Maybe it's as simple as showing in the UI how much of a curation reward you've earned for your vote so far on a post. Right now I think it's like you said - people just see big numbers on popular authors' posts and want to get in on it.

I think there are some programs that do that function. But I just like to see myself in the top 20 - better to be top 10 :)

Hey tiger, you would not believe how many people i have asked for exactly this information, i crown you as hero and genius and will follow you this and next lifes, thanks.

No worries! I had the same questions. Finally decided to go dig out the answers!

Thank you for bringing this close to us tiger ;) good post

Thank you! No problem!

I hope I can still ask a question?

If I correctly I better upvote a post that already has 12 upvotes and $0.04 value then a post with 2 upvotes and $1.50 value. Do I understand this correctly?

Yes, you have that right.

The number of upvotes is irrelevant under the current reward structure. The current post payout and the time since posting (due to the 30 minute reverse auction period) are the key metrics.

But also worth noting that with SBD at $3 the rewards from posting are much higher than curation, since posting can be paid out 50/50 SBD / SP and the system assumes SBD is worth $1 in its calculations, while curation is paid fully in SP so does not benefit from the raised SBD levels. This makes circle-voting / self-voting / vote-selling a more economically attractive use of vote power than hunting for curation rewards. The system is a little bit broken in this respect.

I did not really realize that. To bad for me, since I'm one of the very best curation hunters. But looking at the bigger picture I don't think this is the most important issue that Steemit has. I think there are quite a lot more important things they should fix if they want to give Steemit a real chance to grow bigger and to survive.

Best explanations i've seen so far, thanks alot for your post @miniature-tiger, can't wait to read the next one :)

Steem on ! :D

Thanks! Part 2 should be ready in a couple of days.

WOW, this is amazing :) I upvoted your last post, because it is already to late, to upvote this one! Great job. Added to bookmarks!

Very nice idea with this graphic representation payouts :)

Thank you @noisy! It's nice to see it's still getting some use now the seven days are long past.

And thanks for the upvote of my latest post, that's very thoughtful - I've had a few upvotes on this post over the last few hours!

Epic post on curation buddy!

🥇

Looking forward to to part 2!

Thanks abh! It's taken an age to put all the pieces together. With a bit more research I'll be heading to the top of your curation rankings!

I don't doubt it, very comprehensive!

Let's see next week then 😁

Thanks for this helpful article, and the great explanation of your graphs, @miniature-tiger. Curation is a challenging feature for a newbie like myself to understand. Cheers

No problem! Glad it was useful.

this is the best explanation I have seen on steemit to date about how curation rewards are calculated.
Awesome post

Thanks @paulag. That's really very kind of you to say so, particularly given all your bisteemit experience!
I should have part II available later today or early tomorrow. Stay tuned!

Gonna save here to read it later...

It's a long one!

hahah i'm back!! excellent reading, just what I wanted to know! resteem'd for sure!

You nailed in the visuals btw!

Thanks @andre3301! I'm glad to hear that. I wasn't sure if it would make sense to anyone outside of my head!

Coin Marketplace

STEEM 0.35
TRX 0.12
JST 0.040
BTC 70601.11
ETH 3576.21
USDT 1.00
SBD 4.78