Steemit Update: HF21 Testnet, SPS, EIP, Rewards API, SMTs!

in #steemit4 years ago (edited)

HF21 testnet.jpg

Hardfork 21

Hello Steemians, we have been hard at work adding the changes to Hardfork 21 necessary to add a long term funding mechanism to the Steem Proposal System (a/k/a SteemDAO) as well as the Economic Improvement Proposal (EIP). We are happy to announce that we have completed that work and released the code for Hardfork 21 so that public testing may commence.


The testnet will hardfork this Thursday at 1pm EDT. At that time, witnesses and developers will be able to begin testing all of the changes included in this release. This will have no impact on regular Steem users on sites like


Now that the release candidate for HF21 is out, our blockchain team has turned their attention to continuing development of the Smart Media Tokens protocol!

Rewards API

One of the additions to HF21 involves the Rewards Curve which is used to determine how much Steem a given post should receive from the Rewards Pool based on the votes it has earned and the stake (i.e. Steem Power) backing those votes. The release candidate changes the curve from a linear curve to what a convergent linear curve. @vandeberg (Senior Blockchain Engineer at Steemit) explained the convergent linear curve in this post.

In order to demonstrate the effects of the proposed changes, we created a Rewards API which enables us to simulate the rewards payout with different curves and parameters. We are also releasing this API to the public so that anyone can use it to approximate payouts on the chain at any particular time.

Gain_Loss for Convergent Linear Curve Parameterization.png

The above chart demonstrates 3 different constants fed into the convergent linear rewards curve and their effects on comment payouts. As shown in the charts, 2e12 approaches our current linear rewards at roughly 16 STEEM, which is why we feel it will provide a meaningful and balanced change to Steem’s economic distribution.


@roadscape is working on Communities. There are two aspects to Communities: 1. the backend work in Hivemind, and 2. the frontend work on (Social Condenser). For the backend work, he has been working with community members to develop a Hivemind-based protocol specification that meets the needs of as many Steem developers as possible, not just Steemit Inc. For the frontend work, he has begun the process of combing through the user interface assets that were already developed, evaluating whether the code is still useful, and looking for those components which are reusable across approaches.

Financial Report

Our Managing Director, @elipowell, is currently preparing a financial report that will provide some transparency into our costs (primarily AWS costs) and the revenue we have been able to generate through advertising, so be sure to follow her account (@elipowell) if you’re interested in gaining deeper insight into how Steemit Inc. is operating as a company.

The Steemit Team


This should really be called "The Proposal That Will Have No Impact On Economic Improvement"

i undelegated my steem power, as i will curate steem posts :D
i Waited for this to come apx 2 years, i belive that steem is important because the human imput,
Nothing matters as long as boots dictate the top 10 posts, aslong as not humans are voting we are lost.


Reading off that 2e12 contingent linear rewards curve...

Under HF20 a post receiving $1 in upvotes would give the author $0.75.

Under HF21:

  • The reward pool will be 10% lower to fund the SPS.
  • The CLRC will reduce the payout by a further 40% compared to linear.
  • The payout to the author will then be 50% (rather than 75%).
    I make this: $1 * 90% * (1 - 40%) * 50% = 0.27 under HF21

So broadly: $0.75 under HF20 vs $0.27 under HF21, a reduction of about two-thirds for smaller payouts.

Have I understood that correctly?

Math checks out. Nothing like making use of bid-bots to get above $6.40 completely mandatory for every post.

But don't forget future traf will spread around thousands of tiny whale-votes instead of upvoting himself 10x daily... that has to count for something, right?

I think that there will be plenty of innovation to get around the CLRC issues for smaller payouts and comments. These will mostly rely on aggregation techniques, or as you suggest, various forms of vote-selling.

It's a pain though. The CLRC will just bring additional complexity in return for (as far as I'm concerned) no obvious tangible benefit. The "divisible account" issue could be solved through data analysis and downvotes instead.

Downvotes are still unincentivized and rely on altruism and volunteer work for effort. It is important not to allow that to be attacked by ramping up the necessary effort. Keeping rewards concentrated even if it happens via aggregation techniques as you suggest is necessary to manage the workload required for downvotes. If those aggregation techniques are used in an exploitative manner, they can much more easily be countered with downvotes than 100k or 500k comments scattered all over the place.

upvoted due to troll downvote, not that i necessarily agree with your comment, just wanted that out there.

Yep, and a comment receiving $0.08 because some 4000 SP minnow upvotes it at 100%, that now gets you $0.06 will get you $0.00 after HF21 because we still have dust level cut off at 0.02. Great way to kill social interaction on the platform.

The CLRC introduces a huge market inefficiency. If the 4000SP is delegated to a large account that can aggregate it with other SP then the 4000SP retains close to its $0.08 vote value, e.g. through circular voting / vote selling etc.

A user with 4000SP will be much better off selling those $0.08 upvotes and using the liquid proceeds for social interaction and engagement. I expect innovations to happen pretty quickly under HF21 to facilitate this approach. But it's untidy and introduces another layer of complexity that Steem really does not need.

I really wonder, with that shitty non-scaling convergent linear reward curve in place, whatt continued justification can there be to maintain the dust level treshold on 0.02 ? Move the bugger to 0.001, or at least move it to 0.006 so the same level of influence needed today to rise above dust level stays intact when HF21 hits. HF21 will screw over smaller accounts and hinder platform growth badly enough without the dust level threshold.

The dust threshold has more to do with added processing costs at the blockchain level than anything else. Every payout is a significantly expensive operation in terms of resources so there is a reasonable cutoff below which the costs are deemed to high to justify.

and using the liquid proceeds for social interaction and engagement

That's okay. People sending tips or subscribing for premium benefits or boosting each others content or whatever (as well as earning specialized Steem Engine or SMTs tokens via specific projects and communities) does not have the exploitability and other challenges of a common pool. It probably scales better in many ways.

So broadly: $0.75 under HF20 vs $0.27 under HF21

It's disaster.

It's certainly a very significant change for lower earning authors and smaller SP holders. Both groups are likely to reconsider how they interact on the platform.

With RC limited for newbies, the reward curve too is getting a lot harder. Tsk

Posted using Partiko iOS

I remember when I started and was only earning a few cents per post or comment. It takes time to get anything moving here. I'm really unconvinced that hitting new users so hard makes sense.

Indeed it takes time sir @miniature-tiger.

Posted using Partiko iOS

The RC situation will likely improve somewhat once this is implemented because farming via small comment rewards (some of which already does happen now) will likely stop. I don't know if it will make a huge difference but it will make some difference.

It just feels like this is going to be an utter disaster.

Posted using Partiko Android

The SPS is an important change (although hopefully it could still be funded differently). And from the EIP the downvotes is a worthwhile experiment. There are potential toxicity issues but it's the one change that is likely to significantly move the dial on behaviour. It should be possible to clean up trending at the very least.

50/50 could increase manual curation but there are other avenues that will still offer greater economic incentive so the behavioural impact there is hard to call. It would be interesting to see which orcas / whales are committing to pulling delegations and moving to manual curation to get a feel for the potential upside.

The CLRC seems to add very little benefit even when considered as part of an overall package. It also adds a large market inefficiency which will encourage users to work around the system, to the benefit of middlemen. And when combined with the other changes the impact on new accounts / small earners will be huge. That's the part I'd like to see pulled from the HF.

How would it work if you get larger votes?
Losing 2/3 of my rewards looks scary. It most certanly makes creating content i do less worth it.

In practice there will be more downvotes as there will be a downvoting pool which means less money to bad actors and more money for legit ones. Also bidbots will be less profitable. And the 50/50 curation split means it's now much more worth to curate content than before. So people should move away from bidbots and get back to normal curations.

Ultimately if everything goes well the space should improve a lot and your earnings with it. Would you rather earn 10 steem worth 0.40 cents or 4 steem worth 2$ ?

In the end the value of the coin in terms of market cap will have to be backed up by the value of the content platform, and quite frankly, without somehow onboarding top content creators who now reside on ad revenue sharing platforms, that value isn't going to be all that high. In fact the EIP in its current form is more likely to work against onboarding than for it.

New accounts are going to get screwed over badly by the EIP, what means you will need to be out of your mind even if earning just $20 or $30 a month on ad revenue sharing platforms to move to STEEM after HF21. That's not good, and on the long term that is more likely to push STEEM doen to 0.04 than to keep it at 0.40 or help it grow to 4.00.

While I agree with you on the earning sides, I don't think it will push people away as we'll see a much more sane environment where new users can actually thrive and not be like "most of the sp is in bidbot so I'm not earning anything"

But most of the EIP in fact favours growth of the bid bot economy. Just run a little simulation with a few weeks of old data. The narrative and the actual incentives created by the EIP do t actually match. That is, not unless there is a major shift in how the community considers flagging bid bot users indiscriminately.

Have a look at this poll and it's results to get a grasp of the general reluctance towards using flags (that way).

Just run a little simulation with a few weeks of old data

That method is completely invalid since the entire purpose is to change behavior.

That is exactly why simulation is essential. It shows who ends up getting incentives to change behaviour, and what alternatives of behavior are available.

I really hope HF20 has shown and thought us the dangers of ommitting propper "real data" simulations on a hard fork.

But will Steem increase in price because of this?
Steem is 70 in market cap and not many people care about it or know about it.
If nothing changes creators just lose rewards. That is what i fear. :(
I know i could stand on top my head and play violin upside down hanging by my legs from ceiling and i wouldnt get more votes. (haha that is some circus act)

Why think anyone outside steem will notice what we changed and decide to invest?

The reduction from the SPS (10%) and the 50/50 (33%) combine to 40% when taken together. This will be the base reduction applying to all posts before taking the CLRC into account.

For the CLRC, if your post earns more than a certain amount (16 Steem, i.e. $6.40 - according to the figure in the original post, although I'm still to run the figures on this) then you will be hitting linear and 40% will be the reduction. If you earn more than this break-even level then you will get a slight raise (i.e. less than a 40% reduction), as the reductions made by the CLRC on small value posts will end up being spread over larger value posts.

Better Blockchains that are 100% focused on content creators will be out in 1-2 years ;)

SEC cracking down hard, good luck with liquidity when binance shuts down in sept for USA .

The reward pool will be 10% lower to fund the SPS.

If this is what is actually being proposed, then I would reject this hard fork. There is other existing inflation from which funding can be reallocated for the SPS.

If you have understood correctly then we are fucked up mate. Those small payouts you are talking about are what...less than 1$? Everything less than 3$...maybe less than 5$?

Who will be affected most?

I think that the reduction is approximately:

  • 64% reduction in author payout at $1 post.
  • 54% reduction in author payout at $5 post.

Although clearly these are a combination of the three reduction impacts (SPS, CLRC, 50/50) not just the CLRC.

In my view it shifts the balance of the content side of Steem away from being a social network and towards being a blogging platform. Users who currently gain lots of small payouts, i.e. the engagers and social networkers, will be most affected, as well as new users trying to get a foothold on the platform.

So instead of doing the exact opposite, we are trying what exactly?To hammer the payouts of those who earn pennies and discourage all those which we have desperately been waiting to join us for so long?...maybe we are doing things wrong....? Just saying.

fun days are ahead.
could we please just test it fast and get it over it, so i don't read about it any more and just be happy with my 10-20 cents.

Unfortunately, the comment section under this post is - again - another 'great' example for the damaging effect of flag wars on STEEM. What a devastating impression (potential) newbies (and investors) will have when reading all that?

I think there should be an elected committee with lots of delegated SP (for example from Steemit, Inc.) to be able to discuss, decide about and counter abusive flags if necessary. Only then - in addition - a downvote pool made sense in my opinion.


My only question at this point is: what are the specific, measurable, and timely goalposts by which we can judge whether the EIP has performed the black magic to make everything shiny that its proponents claim, and are we prepared to commit to giving up on it when it inevitably fails them?

It's pretty clear by now that this is going to have to be tried and fail dramatically before anyone thinks about doing anything actually productive, and I just hope someone's thinking about how to identify when that failure has happened.


Yes exactly. We need some defined goals so we know when this has failed. Downvote wars that drive users away? More people powering down than they are currently? Less new users coming aboard? What are the metrics? We likely can't just use the price of steem because it will probably ride a random crypto wide pump at some point.

If they had run some simple simulations with a few weeks of data, this would be crystal clear for everyone. The EIP is going to create incentives that favor the bid bot economy.

  • There should be a factor of about 3x in total influence needed to get a post or comment past dust level, incentifying more medium sized accounts to use bid bots.
  • The ROI for bid bot users will decrease while the ROI for bid bot owners will increase, incentifying a drop in price for bid bot usage.
  • The drop in price for bid bot usage will free up money currently being pumped around, creating a demand for new stake delegated to bid bots.
  • The flag pool will make self up voters (who hardly do harm to the content economy) more vulnerable to getting flagged, creating new insentive to delegate to bid bots.

I think there are a few simple fixes possible for this part:

  • Drop dust level from 0.02 to 0.005
  • Curator/Author buckets as described here

There are other issues with the EIP, that make it very likely to backfire even worse. The other type of backfire isn't that easy to simulate or measure after the fact though as it involves new accounts, platform growth and onboarding.

It is really interesting how Steemit believes it can grow its economy while not adressing inflation in any way and while not adressing the fact that STEEM is losing the battle for top content creators. For a real EIP we desperately need usefull features that burn STEEM and we need STEEM to truly compete with add revenue platforms and pull in those top content creators and their follower base from other platforms.

The EIP is going to backfire in more ways than one. The bid bot economy growth clearly being the most predictable and visible. Lets hope that lacking proper simulations (they wouldn't have gone through with the EIP in its current form if they had actually simulated the effects on the bid bot economy), they at least set evaluation parameters for success and failure for the EIP, so we can look forward to a HF22 in a few months that either reverts or fixes the current EIP implementation.

You were pointed out with utmost brevity by @smooth previously that Garbage in = Garbage out.

So asking for simulations based on trash, what do you expect to discover, or you really think Anything meaningful can be ascertained from trash curation, trash Posting and trash engagement?

what are the specific, measurable, and timely goalposts by which we can judge whether the EIP has performed the black magic to make everything shiny that its proponents claim,

Uhm, there is no one here who also understands Spanish?

Pft dudes! just click on this green text 'spanish' link above to find out a pretty insightful video inside of what was meant as a jolly Carnival post three months ago. And you all will have a pretty close idea about how EIP will perform.

PS. Don't worry!! if you don't speak spanish, there are also subtitles in english & portuguese to get the sobering spanks quite right. };)

what are the specific, measurable, and timely goalposts by which we can judge whether the EIP has performed

@trafalgar is probably the best one to answer this or perhaps he can point to some existing posts to answer it.


I hope to see some detailed posts about how the EIP will actually impact users. Most people don't understand hypothetical curves.

Take a few 'actual' accounts at certain levels of Steem Power (50SP / 500SP / 5000 SP etc) and show what they received from the pool in for example the month of May under HF20 and what they would've earned if in that same month of May HF21 was running. Both in author as in curation rewards - see who is going to be the big winners and the big losers here, and if we want/need these people to be the big winners and losers.

I know it will be way more complex than that and some behaviour will change due to HF21 so the numbers will be different, but we have to start somewhere - or have I missed posts that are already working through real-life examples/impact on users?

nah, drop it let it explode and apologize about how they didn't consider the impact later. :)

Oh shit, sorry, of course we should do that - I'm so naive.


It's all counterfactuals. It's easy enough to run the curves (and the post says they will release an API explicitly for that purpose), but the whole point of the EIP is to change up the incentives enough that some behavioral changes result. Calculating the new curves on the old behavior is only a starting place.

I'm in favor of the SPS but not the EIP. I wish they could be separated out instead of having to accept or reject the whole package.

I think the EIP makes too many assumptions about positive behavioral outcomes, without having enough checks in place for negative behavioral outcomes.

My only hope is that big whales that came out in favor of it... behave in a way that proves that the changes were good.

"Look how much better I'm behaving under EIP! I told you it would work!!" - Future Traf

And of course Steemit Inc.'s simulations have never been off by a factor of ten before.


Calculating the new curves on the old behavior is only a starting place.

Yes, I state that in my comment as well - I'm currently very skeptic about how big of a shift in behaviour we'll see, so in my 'uninformed' mind this 'starting place' might in the end prove to be pretty much reality for 95% of the platform anyway :') (Fake LOL, in reality I'm crying here.)

"Look how much better I'm behaving under EIP! I told you it would work!!" - Future Traf

:') Oh, the joy!

I give "New Traf" a week. Look, you already have all the information you need to know if he is trustworthy, cares about content and is interested in the success of the community.

What more is there to know.

While I do hope I am wrong and these changes bring awesome changes. It is still too much for one hardfork.

Those guys...

They've had months to change their ways to offer proof and say, "SEE! This is how it could work!" All talk, no action.

@kevinwong! I'm looking at you, boy! You better be right or I'll be shoving my fist up your ass!

On the bright side, they've extracted so much SP from this place, we're going to be loving those votes they promised, @whatsup.

Bid bot owners would be big winners, but that should self correct partially in the price of bot bids. Social interaction and new accounts would be the big loser, especially small accounts currently making small amounts of money from minnow and dolphin up-votes of their comments and posts.

The only incentive the EIP truly creates that can be easily seen when simulated is incentive for growth of the bid bot economy.

One important incentive the EIP doesn't create is the incentive for good content creators making a few bucks from add revenue sharing on other platforms, to come and switch platforms in favor of STEEM.

Most likely not, bidbots will most likely recieve a lot of downvotes now that they are free which will make their business model much less sound.

It would, I think, be much more effective and way less disruptive to the content economy to combine a 67/33 split for upvotes with a 33/67 split for downvotes using some sort of curator/aythor buckets, instead having 50/50 for both.

"garbage data in= garbage out"

Simulation won't show anything meaningful considering the current curation, posting and engagement behavior.

Why did you pick 16 Steem as the flatting point for the reward curve?

How many people would this effect? What are the pros vs cons? I'm sure this was all laid out in some private discord meeting with witnesses but I'd love to see a little more detail on the logic behind this just so I know you're not pulling numbers out of your ass and put some thought into it.

Just as they were in the past, these numbers were likely pulled out of their ass. Having the curve flat at 16 STEEM is a joke.

I'm sure this was all laid out in some private discord meeting with witnesses

Well, you are deadly wrong in your suspicions @wakeupnd.

Big honchos never meets in plebeians bars as the rest of us. They just talks and start to pat each other asses when fully drunk in that other exclusive underground club they call Slack. };)

Ya, that sounds about right, lol.

Wus' up? @kawaiicrush? ¿Fat fingers syndrome? ¡Wadda fuck! hahahaha :p

I agree 16 Steem is low and a bad idea. Probably should be much higher considering the price of Steem is likely to go much lower.

That's a good point on the price. Sure would look dumb if Steem was at it's all time low of around 11 cents and users could only make a max of $1.76 per post. Part of the Steemit experience is putting out great content with the hopes of getting upvoted by a whale and making $25 bucks or more. I think this curve would end that fun experience for users.

Seems the graph is incorrect because it does not factor in the dust level threshold. Or is dust level going to be abandoned when the new curve is activated?

further, there truly doesn't need to be a steep section to the curve. Try a curve of the shape N x S^log(N) instead.

Here the blue line is a line close to the blue line proposed above. The orange line would be a more fair alternative that pretty much maps to a 17% increase of influence per increase in fish size. Because there are no extra steap or less steep parts in the graph, incentive scales perfectly between all fish sizes, so also there should be no sweet spot for abusers (or bid bot users) to aim for.

$0.75 under HF20 vs $0.27 under HF21

Rip Steemit, reject EIP please!

Coin Marketplace

STEEM 0.25
TRX 0.10
JST 0.032
BTC 41349.04
ETH 2203.93
USDT 1.00
SBD 4.71