Math of Steem: Early votes have more value because 50/50 payout isn't really 50/50

in #steemit7 years ago (edited)

This is the post where I tell you why basically all of the math everyone uses for Steem posts is wrong. Is that enough of a clickbait opener?

So, we have 50/50 post payout as the default, and SBD is over $1 USD so just about everyone is using it. And we like to think about it paying out 50% SBD and 50% Steem Power, and all of our math assumes it pays out that way. Unfortunately, it doesn't.

The problem is curation. Curation, no matter what your choices for payout of the post, always pays out exclusively in Steem Power. Which might make you think that a 50/50 post actually pays out 37.5% in SBD and 62.5% in Steem Power. That's what I thought at first too, and I had a giant moment of

oh crap all our math is wrong

But it turns out it's more complicated than that, and in a way that means that while all of our math is still wrong, a lot of it isn't wrong as meaningfully as it could have been. Some big things that could have been very wrong, like vote-bot profitability calculations, actually turn out to be fine because of the next point.

So I started looking into how votes in the first half-hour actually affect the curation of a post. We've been told that early voting gives some of the voter's curation to the author, but there's not a lot of information about how that happens. And I haven't figured out the algorithm for that yet (maybe in a future post) but I do know the methodology, and it should be a key insight for your voting strategy.

Early votes increase the percentage of post payout that goes to author rewards.

And while curation is still paid out in SP, author rewards on a 50/50 post are paid out 50/50, no matter how large they are. If SBD were still pegged to USD this wouldn't be a big deal. If the peg ever manages to come back, your vote will be worth the same no matter when you cast it. But for the moment, a vote cast in the first half-hour of a post has more value than one cast later. let me repeat that in big text:

Your vote is worth more the earlier you cast it!

In the first half-hour of a post, your vote has more overall value than it does later, and the earlier the better. You won't get as much curation value, but the author of the post will gain substantially more in post payout than you will lose in curation, because their reward comes partly in inflated SBD.

This can make a huge difference. Consider our self-voting whale @snobdoucher (thanks to @meno for the name) who has a $50 vote and knows nobody else here, he just sits in a corner voting his own posts and nobody else knows he exists. All his posts end up at $50, and he's the only one who votes, so he's the only one who gets curation. You might think it shouldn't matter when the vote happens.

Well, if he votes his post instantly, all of that value goes to author rewards and he makes 25 SBD and $25 USD worth of SP, currently worth about $83 USD. But if he votes his post after half an hour, 25% of it is calculated as curation, so he gets 18.75 SBD and $31.25 USD in SP, currently worth about $73.40. That's around a 13% difference in post value, which is huge. This is an example created to maximize the difference, but even if the real effect on your posts is only four or five percent, that can be a big deal.

And that's with SBD at $2.25. If SBD ever goes way up again, it magnifies the effect.

So what does this mean to me?

On a very basic level, it means that if your goal is to support an author, you should vote earlier. It means that looking into voting automatically with services like SteemAuto has a significant amount of value; I've delegated 1500 SP to a separate account @doctorworm whose mission is to vote small authors I want to support on minute zero.

It also means that if you're manually curating and using your rewards to grow your account you should be self-voting your posts, even if it feels icky. You'll always be there at minute zero, and that self-vote will give more value to your future votes than one given to a post made more than thirty minutes ago will give to that user.

It means that you can't tell what the USD value of a post is by looking at the little number at the bottom, or the value of a vote by looking at its number in SteemWorld or Gina or wherever. I'm sure @transparencybot will eventually come along and comment on this post with the nominal value of my Minnowbooster vote, but it will be wrong. (As it points out in the comments, I was incorrect about this.) It's still possible to calculate the USD value, but you need to know the times and sizes of all of the votes already on the post.

To get all Chicago about it:

VOTE EARLY! VOTE OFTEN!

Sort:  

very interesting, I don't understand how STEEM works, but this is a start!

Thanks for this article that further defines the voting and curating during the 1st half hour.

Just wanted to let you know that as of the start of this week, @trasparencybot does calculate the curation based on the exact time elapsed between post creation and each vote, so the numbers are no longer estimated but instead calculated based on the current values of SBD, Steem, rshares, rewards pool and recent claims, all of which of course are very dynamic.

I should have doublechecked that against more recent comments. The one here does seem to be accurate; I don't suppose you'd care to point me at the algorithm?

I've corrected the post.

Thanks for the correction. I had wanted to get to it earlier, but as you know, there is always more to do in a day then time allows.

Of note: The author portion is linear (15 minutes = 50%), but the split of the curation between curators is sqrt, with the earliest voters getting more "help", but the sheer voting SP by the bots almost always takes the cake.

Here is the author portion simplified.

https://steemit.com/utopian-io/@steempytutorials/part-12-how-to-estimate-curation-rewards

Great post. I’ll now promptly self vote this comment!
Seriously thanks for laying out the math.

I was going to upvote this but then I realize you are on my autoupvote list.

Just when I thought I have it all figured out, you unravel something I'll be confused about more. LOL.

It's definitely worth thinking about moving your auto-upvotes forward. Minute zero is the most effective but every minute before 30 is meaningful.

Not necessarily on my posts, I know a lot of my autovoters are here for the curation rewards and I don't have a problem with that.

Okay so if I want to give 100% of a vote to someone and not get any curation at all, vote at 0. Same with self-voting? If I want to give 100% of my self-vote to the curators of my post, vote when?

Okay so if I want to give 100% of a vote to someone and not get any curation at all, vote at 0.

Well, sort of. Even with SteemAuto at minute zero I'm finding I get a tiny bit of curation. But you're also giving more than 100% of your vote to the post author because of the 50/50 difference and the broken SBD peg. Self-votes are the same.

If I want to give 100% of my self-vote to the curators of my post, vote when?

You can only ever give up to 25% of your self-vote to the curators (and it's smaller if you've had immediate votes). To give the maximum possible just be the one to upvote last.

Okay thanks for the information. Will test different things.

Thanks for your very informative articles the only problem for me on a morale not I will never upvote myself it just feels like it’s against what this community is trying to build on. With that said though I’m not going to stop people from it either.

Very informative and fascinating article. Persoanlly, I think we need to hardfork out self-voting and upvoting bots, and the sooner the better, but for the time being your article is importation information for people to know. Still, I will never self upvote, and I will only ever upvote quality even if upvoting crap would benefit me...timing, or no timing... Thanks again. I learned a lot from this post and I do appreciate it very much!

I think one of the implications of any blockchain system is that it's never going to be possible to stop anyone from automating anything.

Maybe not. But hopefully we can find ways to de-incentivize it.

Unless something else comes up, figuring out how to do that with SMTs is going to be the topic of my next #steemit post.

I'll be looking forward to it!

I look forward to that post

Steemit is very complicated for noobies - which still consider myself as one. I enjoy learning about these intracasies so thanks for yoour post.

What about payouts on comments? I read that upvoting, if not 2c or more, is cancelled. Also, does the author of the comment get the full payment (50%-50% if set up that way) or is there a curating or other kind of deduction?

Comments are paid out exactly the same way as posts. So since I happened to see this and upvote it in minute one you should get a nice tiny bump.

I read that upvoting, if not 2c or more, is cancelled.

If the total payout is less than 2c it doesn't pay out. It doesn't necessarily mean a smaller upvote won't, it just needs to have more votes on the post/comment to make it work. There's now a service @dustsweeper that can clean those up and make they they pay out, as well.

This post has received votes totaling more than $50.00 from the following pay for vote services:

minnowbooster upvote in the amount of $213.28 STU, $296.74 USD.

For a total calculated value of $213 STU, $297 USD before curation, with a calculated curation of $53 USD.

This information is being presented in the interest of transparency on our platform and is by no means a judgement as to the quality of this post.